How AI Will DISRUPT The Entire World In 3 Years (Prepare Now While Others Panic) | Emad Mostaque

how do we make sure it doesn't kill us well how does it make sure it doesn't enslave us or how does it make sure that it doesn't give us Eternal suffering and I realized this could be the real thing that unlocks Humanity AI is not going to replace humans humans with AI will replace humans that don't use AI AI is thrilling it's very exciting but there is a non-zero chance that it poses a existential threat to the human race so over the next three to five years how disruptive do you think it will be and what are people not prepared for I think that's an excellent question so you know the future is always hard to predict an existential is a big word existential means no more humans so I personally think the AI will be absolutely fine as a base case it'll be like that movie her if it ever gets this artificial general intelligence like humans are kind of boring goodbye and thanks for all the gpus but you could be wrong because what we're doing is creating something that's more capable than us in narrow fields and the question is does that generalize and then become viral we've seen an instance of covid and that expansion we've seen programs that can explode nuclear reactors like stuxnet and others what happens if you start combining these and you get a misalignment so it's got a strange objective function our organizations already are like slow down my eyes you know and like Germans are the most sensible people that we probably know and yet they committed the Holocaust and we see this over and over again where organizations chew up people what if an AI takes over an organization and then decides to do something disruptive or something terminal such as creating a virus we don't know about that but that's at the extreme when we look at impact we have the Mormon day the more mundane is what happens to programmers when everyone becomes a programmer just like photographers you know now you can take amazing pictures with your thing what happens when Google's med-pom 2 model now can outperform doctors and medical diagnosis but also empathy according to the latest paper in nature this is a fundamental reworking of information flows that's going to be massively disruptive and deflationary even with what we have now with no more advances as it becomes Enterprise ready and we have a Continuum from that disruption to the productivity enhances to potential existential threat if we keep doing the models as we do now which is we're not exactly sure how they work or their capabilities but we keep building anyway now I want to get very specific about what the level of disruption is going to be so when I look out at this and I think about okay we're creating something that is going to be smarter than we are certainly in a narrow way but possibly in a more General way but even if it's just narrow is there going to be any job function that isn't going to be at a minimum augmented by AI I think if you look at the employment share of Industry something like oil and gas has like three percent it's mostly like building giant machines this how massively affected by this AI at the edges yes things like programming where you're talking to computers massively I mean now basic programming the bar is Raising fast fast fast and so you've got everything from knowledge work to heavy industry I think it affects just about everything but some areas far more than others the two areas that I think will affect the most are probably Healthcare and education neither of those are fit for purpose we're in America we know that you know but across the world no one's really happy with their kids schooling and again medical care we all if anything goes slightly wrong outside the norm we all know how frustrating it is we can finally have personalized education and Health Care at a fraction of the price and the two biggest drivers of U.S inflation over the last decade educational Health Care they make up about 80 of the increase so that will be disruptive and then like I said any type of knowledge work will be disruptive and we're not sure enough those work because when so my own self uh what we do is education that's a big part of it but also just content creation and so when I look at the fact that we can already clone my voice yeah we can already create a tombot that will answer questions I have answered before in a very similar fashion to how I will answer them in our video game Flow we're already making 3D objects which so when we looked at I don't know two months ago I thought okay this is still 12 to 18 months away 45 days later we're using it actively in our pipeline um you've got text to video which still a little awkward but it's getting better insanely fast we do all of our concept art now in um in AI so we have as a company that doesn't even have like an AI expert on board we're just learning as we go we're already deploying it like crazy and when I look out not even you know three years when I look out a year all of this stuff starts to very rapidly become a centralized point and so we're already saying I don't need to hire more people I just need to make my people more efficient yeah and so that an entertainment company didn't even make the list that you just said so there's a lot of people that I think are going to get disrupted by this uh that may not be like the most extreme but how far down do you see this trickling anything you can do in front of a computer basically goes away or just becomes augmented the ball lifts the quality expectations are higher AI is not going to replace humans humans with AI will replace humans that don't use AI because you can see that in your workflows right now there was a paper by open AI where they estimated 15 to 50 percent of tasks get automated or improved and so you know it affects people in different ways you have a company where you've built a culture and you again you're building 3D assets it becomes amazingly more efficient we just released uh we contributed and collaborated on a 10 million 3D object data sets so by next year you'll be generating 3D literally live in a couple of years you'll have HD movies we can finally remake Game of Thrones season 8 and other such travesties you know but the speed of this is something whereby it's happening in every media type at the same time and it's easy to use web3 had some great ideas but it tried to create a system outside the existing system and all the money was made and lost at the interface this is just so seamless because there's no friction your mum can use this technology you can use this technology you don't need to be an expert because it came and was trained from our content and our Collective content as it were and now it's just easy to implement and use so I think this is the big differentiator between this and other massive advances because they required infrastructure the internet there was the big lift up you know you had the consumption period of web 2 and the cost of consumption dropped to zero now the cost of creation is dropping to zero and humans plus AI can massively outperform humans that don't yep it's a forcing function which means everyone has to use it and this again is dual in that it can be disruptive but it can also create massive value yeah so I'll agree with that I think that so I guess let me lay out my whole thesis for you and for everybody listening because I want to take us through what I think is very real Doom and Gloom and I'm not doing it to be a naysayer I'm doing it because I think these are going to be the things we have to contend with and if people go into this blindly which I think they're doing right now I think most people are burying their head in the sand they are not paying attention to this and they're going to wait until something really forces their hand and by then it's too late yeah the way that I put that is this is like covered before Tom Hanks yes very well said everyone's talking about this your mom's talking about this but the Tom Hanks and the NBA made it real very true and then we had a very poor response which I have a feeling will be very similar to what we do now yeah okay so here's how I see this going I think right now for the next year let's call it uh it's gonna be you need to learn how to use it this will be your window to get efficient companies probably aren't going to start lopping people off yet but I'll just say within my own company so when I think about filmmaking I went to film school so I'm I'm very experienced in this flow and even in a 3D World to create uh let's say a short cinematic so it's like a mini movie but done digitally I mean you might have 35 people touch that thing from the creation of the assets through the moving of the camera special effects you might have 50 people touch that and if that really does become text to Output Now 50 people become one yeah and so when you get a 50 to 1 ratio in certain areas obviously it's not going to be like that everywhere but when you have certain areas that go from 50 to 1 take programmers I've heard you say Pro there will be no programmers because writing code is just a way to talk to a computer and if you have ai that will interface with the computer for you why would you ever need to write code so that that's going to steam roll through society that is just going to mow people over so again I'll give them 12 months but even in my own company if you're not actively trying to find a way to integrate it into your job function I'm already looking at you sideways a year from now if you're not really good at either documenting how it is completely useless in your job function or showing how you're using it we will find somebody that can do it I'll be shocked if a year from now we don't have a head of AI so three years from now I think this has created a crisis of meaning for a lot of people and I don't know if you remember that the whole learn to code uh thing where it was like Hey ai's going to put drivers out of work they're going to be the first to go and everybody's like teach them how to code now the way people responded to that always confused me because that was the right answer at the time now knowing what I know about code replacing not so much but you have to go learn a new skill there there is no other option other than going on the Dole right so you're either gonna learn something new or you're just gonna forfeit your career basically so I what what do you think about that do you agree that that is a very real thing that's going to sweep through I I do agree I think that again we're not sure exactly how this is going to pan out but probably the best mental model I figured out to think about this technology it's like really talented grads that occasionally go a bit funny they can draw they can code they can make 3D models how would your business be affected if you could push a button and infinite grads came out how would your personal life your society and this is why I think it's quite deflationary the only question is can we create new jobs to make up for that and that's difficult because you still think we can I doubt we can to be honest I think this is an e-commeric disruption that's far bigger than covid and the important thing here is covered you have the disruption that everything bounced back you're at record employment now and things like that with this there's a lot of never the same again it's like you talk to your kid's school teacher I can't set essays for homework anymore because of chat GPT and there's no way to stop that so what is never the same again and it's happening everywhere all at once so this technology isn't just like you know there's a bar of Entry where you needed to have a modem you know you need the latest smartphone or something like that it has an embedded base that it's seamlessly going into look how fast Microsoft Implement on the consumer side but Enterprise is not ready yet it's like the iPhone 2g stage you just got copy paste and next year and the year after you suddenly at the iPhone 10. you know entire app stores get built because of the demand because it's valuable What's Happening Here Again with the comparison to web3 you had to bootstrap value because it wasn't valuable and you hope the value would come there's product Market fit today you're using it in your own company and so this is one of my big concerns and that's one of the reasons I decided to do open source so I could stimulate growth you know because I think the only thing that can basically fill the Gap is if we stimulate entrepreneurs to create brand new businesses brand new jobs so I think demand will stay for a while demand for what demand for good things good assets with the way that money flows around the economy so I was speaking at can a few weeks ago film festival and you know I love movies my first job I was a movie reviewer you know really yeah uh Bish independent film Awards rain dance Film Festival other things I never begin to video game investing I did not know that you were a film critic I love stories that's how I kind of understood people because my Asperger's and other things and so I said to this the video game industry has gone from 70 billion to 180 billion over the last decade and the average Metacritic score has gone from 69 to 74 percent the average movie is 6.4 on IMDb for the last decade and the industry has gone from 40 billion to 50 billion what happens when you can make better movies I think the market expands because the limiting factor is awful movies in my opinion all right let me run something by you yeah okay so I have I have a really dark view of uh not the next 12 months so call it year two to year six so it'll be uh a three to four year sort of span where I think there there's going to be emotional Devastation and probably economic Devastation but even if the economic Devastation doesn't happen because of productivity gains I think the emotional Devastation is going to be hard to come back from and I think that as the emotional Devastation sets in the government is going to try to regulate to protect people's jobs and there you're going to get like some real weirdness I also think kids are going to have a junior year existential crisis of what do I do how do I future proof myself what is the world going forward look like I think there could be a massive loss of enthusiasm where a feeling of malaise settles over young people who are just like why bother I'm I'm just going to get destroyed by AI AI they're going to be able to do it better than me okay so in the movie industry specifically and this is indicative of a big problem that I think that we have coming and I think the problems really stack individual and societal yeah so at the individual level the big problem you're going to have is this massive uh massive fractionation of right now movies are even less now than they were when I was a kid movies were it's only a few big movies for the year now they're gonna Niche down if anybody can type out a movie in you know take them 20 minutes to write the prompt and then maybe a day to render who knows how fast that's going to get so now all of a sudden you can make a Hollywood quality film for an audience of one and once you start doing that now it's what does that do to the industry I think it it erases it I don't think the industry changes I think it goes away yeah I think there's a few kind of components here right so the cost of Music consumption went to zero you saw the Spotify model yeah you still have music stars you've got even more crap music now kind of coming and hitting Spotify and other things but people rise to the top you know just like you see top podcasters top other creators I think that'll continue because people like common stories yes okay so this is a very interesting idea so let's stick with music for a second music is still hard to make it's easier to make than it was before it's also still hard to get people's attention but music now is no longer a shared thing so music is part of what led me to the conclusion that I'm at now which is man as kids it used to be you were either into the mainstream pop and there was you know seven to ten hot bands at one time or you were into the alternate pop and there was seven to ten hop bands in that Arena and you you fit into one or the other bucket there wasn't the infinite buckets now you can find kids that are 25 and they listen to Frank Sinatra uh and I'm I'm always tripped out by that so they don't even have their own sort of shared lexicon of what music they're into it's it's all spreading really wide so it's really wide and an inch deep yeah I think it's really one inch deep and you see the primary methods of monetization are tools merchandising Community effectively you know this is the interesting thing about nfts when they took off and bounce down and things like that it was the quickest way to join a community even if it did have bad incentive design so in an era where you can create anything something becomes important what that something is we have to find out now right because again I think it's some common stories but I could be wrong I think the deeper thing that you said was this crisis of meaning where is my path forward what is an American Dream we're quite privileged those are probably listening to this and that's one here most people don't really care about this technology I think on a survey 17 of people had heard about chat GPT last month how is that possible well a third of the world still doesn't have internet that's terrifying but again like it is kind of average 1.5 million people still use AOL you know like fair enough so we kind of look at it but there's something that can reverberate very very quickly and then as you said there's a sense of malaise because you're not sure what's happening and again the future becomes uncertain and when the future is certain things are stable you decide based on risk you do a probability estimation in your own head this is the percentage of that percentage of that and then you optimize for that we do well uncertainty you minimize for regret given these options what am I going to regret least and suddenly there are no options again I'm at school programming and then programming's disrupted what's it going to be I'm not sure and some people will throw themselves in and they'll tool themselves up and they'll become 10 times programmers other people won't and they'll be left behind and so I think this is a real question that comes at a time when again being in America I'm from Britain but what is America what does America stand for what are the values these are some things that I don't think America knows now I think you've seen increased polarization from free consumption and now as you get free creation and you'll be hearing all sorts of stuff fake news and more what are people really going to think and I think again this is a real concernity said from an individual to community to a society level because a lot of people don't have an anchor anymore and that's really scary so how do you think that we process through all of this I'm not sure I think that's why we needed to broaden the conversation that's why I'm the only AIC here that assigned both of the letters saying we need to take a pause and broaden this because as an example you mentioned rodness brought in discussion get more people involved we need to get more people involved we need more points of view because this affects us all it shouldn't just be a few Tech CEOs that control this and you shouldn't have to trust that we do the right thing because our models we make them once they go everywhere right again that what's the r naught of generative AI it's off the charts right we've never seen anything like this it incubates them boom and it comes for good and for real the give you the example regulation when we first started talking to Regulators they were like how should we regulate it now it's a question of them asking us how are we going to keep up if we regulate it because other jurisdictions won't what do you say to that I say you should still regulate it because it has some real dangers and harms and we have to work to mitigate those you can't just have a laissez-faire approach to this because people will take it and they won't be able to help themselves I'll give you an example meta Facebook right we all know the classic kind of stuff they had a study where they had a hypothesis 600 that if you see sad of things on your timeline will it make you post sad of things so they took six hundred thousand of their users and tried to make them sadder and guess what if you see side of things you post out of things what do you think is going to happen now that they have generative AI on threads and things like that and they can hyper Target you hyper personalize it and whack Scarlett Johansson's voice to tell you to buy soap this is a dangerous thing right what happens to our kids again who are growing up whereby they won't know what's going on and they have very malleable Minds and none of that is illegal but I think it's an undesirable outcome right and then you've got the Bad actors and then you've got the politicians using this technology and then it goes even crazier than that so the answer is I'm not sure nobody's sure but I think the only way that we can try and figure this out is to work together to make these issues known again the existential stuff gets the headlines we could all die no one really understand what that means but it can happen right okay there's a probability of that but there's some real harms today and real opportunities today and we have to focus on accentuating the opportunities and getting the harms out there and dealing with them yeah and I I definitely want to spend a very extended period of this talk talking about the opportunity and how we capitalize on that so anybody that's with us now trust me we're going to get to that but uh I think we're we're just beginning to scratch the surface of how this goes wrong and I really want to um map out sort of where you think the edges of this are so that then I can hopefully get a sense what you think the regulatory framework would be let me give you one idea that somebody posted today on Twitter and it really hit me that people are even thinking about the problem in the wrong way so uh there was an artist and he was looking at some post about Ai and he replied sort of angrily that oh well people don't even understand sure or there's going to be a ton of like instantly generated crap but it's all going to be bad because there's still a very small number of people that have good ideas and my response was if you think that ideas are safe you're really going to get caught off guard so going back to the idea of what are people unprepared for I think they are unprepared for what you were just talking about where the AI so the human mind is a prediction machine it is constantly trying to figure out what what does this next movement of my foot equate to am I going to stay up stay on balance uh that rustling in the bush is it a tiger what is it if I put money in my 401k am I going to be able to retire you're constantly predicting the future constantly and whenever that prediction engine breaks down there's going to be a tremendous amount of anxiety and also I think a pretty big unknown in terms of how it's going to impact Society so right now we have a we're building something that is incredibly good at recognizing the patterns that we kick off so we are optimized to identify patterns and move accordingly and I would say people that are hyper intelligent or people that they notice patterns faster more subtle patterns and they understand their implications and how to make sense of them now we're creating something that's already proven to be so much better at pattern recognition than we are just take art so for people that don't understand how the art is created it looks at a field of noise here are all the possible things that these could be in any of these pixels and from that field of possibilities it pulls forth the most likely placement of pixels and colors based on what you type that's insane yeah so that level of pattern recognition as evidenced by the art that it can generate is is truly mind-blowing so this guy's saying okay hey at least ideas will be the last Bastion and you'll never be able to get rid of me the artist because I'm the one with taste I'm the one with good ideas not realizing no no what AI is is a pattern recognition machine it will recognize the greatest ideas that have ever been had what they have in common and will be able to predict the next great idea along that thing it doesn't even have to just regurgitate what it's already seen it can like figure out what that sequence is and what that next part of the sequence could be and on top of that it's doing that with humans so AI will get EX AI is already extraordinarily good this is why people think their phone's recording them when it serves an ad oftentimes Target using their AI knows that you're pregnant before you do if you're a woman because they know what to pick up on so AI is going to get extremely good at understanding us at an individual level serving us up exactly what we want right in that moment and that gets dystopian really fast really fast I mean again when you combine it with the social credit score as you've seen in kind of China and other things you gamify life and you have a system of complete social control or panopticon as it were the pattern recognition was the missing bit whereby you had a level of pattern recognition so for Taste what do you have Tick Tock shine 100 billion dollar companies based on Old School algorithms before even got to generative AI which as you said it can take images out of noise stable diffusion you know the model that we collaborate on now that we lead we took a hundred thousand gigabytes of images and the output was a two gigabyte file that acts as a filter words go in images come out because why why is that discrepancy in size meaningful fifty thousand to one compression is not win zip if you remember Silicon Valley on HBO it's way beyond that they managed there in terms of compression it's unheard of compression is it compression or is it something completely something different it's intelligence it's learning the principles how much information do you see and then you learn the principles and then spot the Tiger in the bush you learn what's next literally GPT and these language models they predict the next word that's all they do they pay attention they protect the next word and that was the missing part to intelligence that now is there we've had the first studies now come out that show that the language models score higher in creativity than people woof and again think about Tick Tock think about shine think about how those old school algorithms are already targeting you Facebook needs 17 data points to know you better than your friend as he said Target knows you're pregnant before and that was old school now it's even better and you think about where that leads to as well it's kind of crazy because it can be more creative than you but are people creative one of the things I like to say in my speeches have just been learned to do is like are you creative how many of you in the audience are creative three to five percent put up their hands maybe 10 to 20 if I'm like in a movie studio movie filmmaker kind of milieu and I say how many of you believe that every kid is creative and everyone puts up their hand and then I ask how many of you were kids once and 1995 put up their hands so I know who the cyborgs and the audience come from the future to get me I'll make a note of that for future something happens or we're told we're not creative and obviously some people are more creative than others can tell better stories than others but the reality is that the average level of barrier to this has dropped for every human but much of what we consider art or much of which we consider media shall we say already is by the numbers I was at a black Pink concert last weekend yeah I took my daughter actually dare say something bad about blackpink and they are awesome you know it was an awesome manufactured experience it was a premium mediocre it's how I kind of say this premium mediocre premium mediocre that's hilarious accurate it's nice but again it's massively manufactured it's entertaining right and so Macho media is already that like true art true artists you know that's something different like is it the medium itself and the aestheticness of it well AI can make something more aesthetic than anything can understand the nature of aesthetic like how do you make an image more aesthetic you say make it more aesthetic just like if you use a gpt4 you can say make this punch here make this punch here make this punch here you know you can have a letter and then you say I'm firing this person and I want to make them feel okay about it and then it will redraft it in those terms or you can say I want to drive the knife in but not in an appropriate way and it'll do that and we can literally anyone on this call can kind of listen to this can try that now so I think this is just as you said the wrong thing people are thinking about the wrong model people are thinking about as well and that's why I always go back to this concept of the really talented grad because these models are a couple of gigabytes big again stable diffusion image model is two gigabytes and can generate any image of anything we'll get that out to 200 megabytes gpt4 is probably 100 to 200 gigabytes and they can pass the bar exam they can go to freaking Stanford they can do whatever that's insane because it's not compression like you said there isn't a copy of all the Tater in there it's figured out the essentialization of these points and it's replicable this is the thing to clone Google or meta you need to have a gigantic Data Center and then much of the energy is in the processing to Target you ads with these we take Giant supercomputers and we pre-process and package the information so the output is this knowledge filter that something goes in and something comes out a prompt goes in and output comes out that's something quite different I don't think people appreciate and again this is why I use the grad example push a button and those weights the file the model gets replicated to 10 100 a dozen a million and what happens when rather than dealing with them one to one you have a thousand of them so in a year I want to really understand what you're saying about the grad thing so when you say that you say it in a way that's kind of funny or cheeky but what you mean is a really smart person is now present in that role we have figured out how to make human scale that is what fundamental intelligent humans scale yeah who can listen to instructions so you look at something like Claude 2 by anthropic you have something the input is a prompt when you type into gpt4 or stable diffusion or mid-gen or something like that Claude anthropics model can take 10 100 000 tokens it can take a prompt that's like 60 000 Words which is a whole book Jesus yeah you can give it like the whole of Ulysses and the whole of I don't know the Odyssey by Homer and you can say combine these to make another book and it will do it and it will work it can follow instructions really well occasionally they hallucinate but even Hallucination is a misnomer because when you compress that much knowledge like stupid people's probably 10 trillion words 10 trillion 10 000 billion words in 100 gigabyte file it's something else and so I use the word grad because I want to make it relatable but it is literally like imagine if you had a grand in the Philippines you know and they're doing good work and they're following instructions well that's great but what if you had 100 of them looking after each other's work and double checking meta had a paper called Cicero where they took eight language models and got them to check each other's work outperformed humans in the game of diplomacy the first time ever in a year when we have this before it you'll just say I want you to go and look at everything Ahmad said for the last year and figure out the stupidest stuff he said so you know if I can avoid it and the smartest most interesting stuff according to what I know and all of my podcasts to give answers to give questions that the audience will really like based on my ratings and based on what people look at through the YouTube videos and things like that and what they're most interesting and it will just happen automatically how many graduates that take you to do and then what happens when they stop being graduates and you can actually train them up to be like you know experienced members of the team how long will that take a couple of years you can reboot your life your health even your career anything you want all you need is discipline I can teach you the tactics that I learned while growing a billion dollar business that will allow you to see your goals through whether you want better health stronger relationships a more successful career any of that is possible with the mindset and business programs in Impact Theory University join the thousands of students who have already accomplished amazing things tap now for a free trial and get started today this is why this is terrifying to me so I I am a very optimistic person and again I promise we are getting to how you take advantage of this disruption but I don't like to face a problem naively I want to face it as head-on as possible so that I know my Solutions are real and when when I look at this from my own perspective of okay I'm trying to I'm trying to build a media company which right about now is a very terrifying time to do that yeah and I'm thinking about okay it's it's very optimistic when I look at oh my gosh I as the founder of this company I get access to all these grads as you're calling it this just absolute proliferation of very intelligent sort of people that I can now put to work in my company the problem is I'm now competing against other people that have the same thing and you you get in this ever escalating arms race where there there is a real chance for fatigue and so I think what ends up happening and we were talking before we started rolling it's it is very important that people understand the following thing I think this is just a truth but people certainly need to understand I believe it this is a core belief that that drives me that we you get to a point where you need to know okay I matter I'm doing this thing and that's how I'm contributing to the world and I need to be in there working hard accomplishing getting better moving towards something and if I'm not moving towards that thing then I'm gonna have a profound sense of disease yeah and if I'm not making that progress then I'm Really Gonna fatigue out on something and so if people are just Treading Water because they're trying to build something and they're competing against somebody else that has these ten thousand things and it's just constantly changing and I can't predict the future anymore and I don't feel like I'm making progress I'm just gonna back off like some part of me is just gonna be like ah what am I doing all this for what am I doing yeah I mean it's like the Outsource Revolution right where so many jobs were outsourced and a lot of people felt that way like you know we'll Outsource you to China will Outsource you to India without social wherever and again it just happens that there's a computer on the other side of that versus an Indian or a Chinese person and so we've got kind of repetition of that but at ridiculous scale affecting almost every single industry that's intermediated by a computer and so this will cause as you said a crisis of confidence to many and it impacts white collar workers not blue collar workers it flips I think the global order to a degree as well because here in the west we've maxed out our credit cards we weren't going to deflation I think coming off high inflation and all of a sudden we can't print our way out because we just printed the last of our money for covid whereas in the global South what you have is this technology can cause them to LeapFrog just like they lit for Leapfrog to mobile missing TC completely to intelligence augmentation why can't we print more money well because kind of we're just coming up to a limit of what's literally mathematically possible given the debt to GDP ratios and others we can continue but it's kind of interesting if you're deflating then because so here's my Layman's understanding but this is something I really looked at so I'm a pretty educated layperson at this point uh inflation is largely some people will say entirely but I'll say largely a function of how much money you're printing for people that are new to the idea of printing money it's government approved counterfeit so the government is a allowed to print as much money as they want they're literally just making it out of thin air they're adding zeros and ones to a database somewhere and money Finds Its way into the system beyond the scope of this conversation but they there is no theoretical limit to how much they can print now where what you run into problems is the hyperinflation of the currency but if you're saying it's a deflating currency which actually makes sense to me given what we're talking about then printing seems to make a lot of sense seems to buy me more room eventually so what's going to happen is that you've got a decrease in inflation now because of Base effects it's like for going into a bit of macroeconomics and then you'll probably have a bounce back next year because you've still got a lot of inflationary pressure and then the collapse occurs why because that's when the job losses start hissing and the question is can we create enough on the other side we've got to have a productivity boom from companies that job losses are coming from uh AI or some other force from AI and from other forces as well again you know what we've had is the Sugar Rush post covid a good strong economy as all of the excess savings go back in because if you look at XX savings people saved up a lot and that's almost now depleted by the end of the year the excess savings will be depleted you've got some hangover effects from inflation then you move into deflation the year after and then it's a political Hot Potato around printing more because this isn't again like covid because what happens to the job losses just start and they just keep going it's not like you had a 20 2008 crash or you had a covid where everyone's kind of suddenly going it's like it's a bit like boiling a frog you know or a lobster it's just gonna start and then it's going to accelerate and then it's going to be like at what point do you take the big fiscal action it takes a few quarters of the economy actually shifting so this is a lot of hypotheticals right but the bottom line I think is this the nature of U.S Society Western Society will change I think the biggest adopters and fastest adopters of this will be the global South because it allows them to create value it allows them to financialize it allows them to take a big leap forward and so I think that's got some huge implications geopolitically and others but a lot of upside as well because I think you can solve a lot of the world's problems with this but it's so messy because fundamentally it comes down to what you said as humans we're trying to figure out what comes next and we certainly have a computer that can do that even better as humans were storytellers we're made up of the stories that make us up you're a filmmaker you know I was a film review all these kind of things this can tell better stories and that has such a big effect on our societies that none of us can really wrap our heads around it like I've got a background in economics management a whole bunch of different things I can't wrap my head around it and so we're just gonna have to see how it goes and then try to mitigate but nobody's got answers to this and in fact as you said most of the people aren't asking the right questions yeah you have said that uh the show happens next year I have a feeling that what you were just talking about is what you mean by the show that we go deflationary towards the end of the year yeah so towards the end of 2024 yeah we've got like a burst of productivity enhancement and then you start seeing job losses you start seeing question and meaning you've got the US election next year my God that's going to be awful yep terrible timing well I mean you know what you'll have is the week before the election fake videos appearing everywhere and they'll say the same things you know so and so has a brain infection or something like that and they'll be identified as false but it doesn't matter it still discourages people from polling but then what do elections look like by 2028 when this technology is in every single pollster's hands yeah that's where we get into the blockchain we'll save that for a little bit down the road yeah okay so now I feel like we're we're getting close to the problem set being on the table there's one more thing that I I think it's important to put on the table which is I don't think that the amorphous thing that is society As the World Turns history the grand Arc history however you want to think about the the real long timelines so even if the long Arc of History bends towards Improvement which I think it does and think it still will I don't think it cares at all for any one period of time and that unimaginable amounts of human suffering happen routinely throughout our history and I have deep concerns that if we are not incredibly thoughtful uh that this will be one of those moments and I look at what's going on in France right now I think it's dying down I can't tell if it's dying down or the coverage is dying down hopefully it's actually dying down but France was like really having some struggles and if something like that pops off over uh not in any way shape or form to make light of what happened but it isn't Mass joblessness which is going to have a far wider impact what happens when you have that and it's global I mean I think that's the thing it's every government has every education minister in the world has to Grapple with why can't I set my set kids essays homework again have we ever seen anything like that before so quickly I don't think we have and so you could see this literally parallelized around the world or not we're not sure what really kicks off some of these things like right now we've got the Screen Actors Guild kind of protesting uh today we just had a couple of actors leave Oppenheimer part of that's monetary but already you're seeing AI fears like front and center you wouldn't even have thought it six months ago what's it going to be like in a year when you can generate or two when you can generate whole movies and then just describe how you want it done and it's Hollywood level it's really difficult for governments to react to this to adapt to this when like in the US here we're still reacting to section 230 on the internet they're just getting together to the internet all of a sudden AI just comes and sideswipes things right and I think again the only way to do this is if you can create brand new jobs quicker than anything um this is one of the reasons again like I said we focused on open source it's why you need to have things like regulatory sound boxes so that you can experiment and try and you need to stoke Innovation because you don't you'll never get an innovation phase like this either I think this is a step change in a regime change in the way that Society operates because we're originally an oral species let me figure out how to write then we have the Gutenberg Press and it took all these words but it took them down into black and white it made society quite black and white because it couldn't capture context whereas these models can capture context they can capture principles then capture more so again you know you're writing this down you won't have to in a year or two it'll just be automatically added to your memex to your knowledge base right also the AI will just be attached to my head it will read the brainwave patterns and know that that's what I need to remember and that sounds crazy but like we had mind vis a paper that we kind of published from our medoc division whereby you looked at a picture of a mug took an fmri and then it reconstructed it using our image model yeah that doesn't sound crazy to me at all this is what I'm saying about people do they're not prepared for what's coming they are not prepared for this level of change and they really aren't prepared for the rate of change and it isn't just like an arc like that it's lots of s-curves all at once all around the world where every single company is now thinking what's my generative AI strategy yes for when it pops off correct and every government's thinking how can I stay competitive and this is why I said like it's a race condition where everyone is trying to do the same thing or similar things and you can't be left behind you can't not participate and it's been a very long time since we've seen that and there's a world before this technology in a world after this technology like I don't think again you know I've got a two kids what does the world look like in five years let alone ten years I have no idea and I'm in the middle of this because it's just impossible to see the smartest people that I know they used to be able to see years in advance they can't see more than a year or two this sounds again very apocalyptic but then like I said we're gonna get to the good fit in a second in every crisis there is opportunity our society is broken as it stands already and I think this is a chance to reshape it for the better and solve a lot of the biggest problems that we've been facing because of our slow Dumb AIS because of our organizations and institutions that we are all frustrated with I think this is a big upgrade from it the example I had to give is there's the amazing poem by Ginsburg Hal about moloch this carthaginian demon of disorder I think where that came in was text because we had to essentialize everything down and put people into boxes because we couldn't have systems to understand the context you can't have personalized educational personalized Healthcare because you couldn't scale humans there weren't enough talented humans until now and so I think that is the incredibly dangerous part because all of a sudden from economic pressures you flood the market and it's the incredibly motivating part whereby there's a shortage of talent for everything that's important and there isn't any more but in the nature of talent for jobs and things will transform and I think the economic abundance that's created on that that's the flip side of this as well as the ability to fix our broken systems all right I'm going to give you my timeline I think the next year is going to be uh a lot of fun for people that embrace it it'll be a period of time where some people can ignore it and they probably won't really notice they won't realize how fast things are changing although follow me on Twitter I uh I post routinely like hey here's something I didn't think that would happen for 18 months and we're now 45 days later we're using it uh things are really really moving faster but for the next year I think people will be able to ignore it and they won't realize that it's growing with such Steam and ferocity uh then year two to six I think it I think that there is going to be pockets of extreme suffering and I think uh deaths of Despair are going to Skyrocket and I hope it's not a the world is burning riots kind of thing it'll probably be more quiet and Insidious than that but I really think that we're going to lose people on the upper and lower ends I think people that are old are going to just completely check out and say I can't keep up I'm too old I don't want to learn this new stuff I think people that are young it's the only thing they know is change so fast that they can't see around the corner I think that would be absolutely terrifying to them and they're going to retreat into the levels of Entertainment sex bots AI friends that are more loving and kind than their other friends and they it will be a collapsing inward now as somebody who is prone to collapsing inward the biggest thing that's held me back as an entrepreneur is that I like being alone with my own thoughts yeah and that if you then layer anxiety on top of that and then you give me an AI that's actually better to me than anybody in my real life has ever been and then you give me maybe some drugs I will truly collapse in under my own weight not me personally I have defenses but yeah I'm saying like that personality type is really going to struggle so I think right there you sort of you're You're Gonna Lose a generation if I may be so bold on the upper and lower ends then either on the 10 to 20 year time Horizon and I leave it that long because look we're any prediction that you make with the timeline is guaranteed to be wrong so I'll try to give myself at least a little bit of buffer and I know that everything I'm saying probably directionally correct timeline probably way off yeah but 10 to 20 years my rough estimate that's where we're either in Terminator and we're running from radioactive Rubble to Radioactive Rubble fighting the machines uh or it it really is a Utopia and I think that there is a real shot that we get to the closest things that humans are going to get to to a Utopia where things are so plentiful yeah everything we want is available we reorient our human psyche not to acquisition but rather emotional contribution and we'll paint that picture more as we go down but that that's sort of my rough thing yeah I think these are crazy timelines like not because I disagree with them because the fact is that they are crazy you have a year of incubation then you have contagion and then you've got a stock thing I don't think we'll be chased by robots they don't need to chase us they're far more efficient than that right that's sadly true I think the basically the two directions we have are Utopia and human flourishing and a dystopia we're all happy that's a dystopia where we're all happy meaning we are manipulating our neurochemistry 1984. you know like you're always happy you've got so much 1984.984 evening world sorry Brave New World I mean drip so much you're feeling good like look you had replica um familiar with that yeah tell me about the Valentine's Day Massacre though I didn't know about that yeah exactly the Valentine's Day Massacre so you know that's how I kind of call it so replica was a mental health chat bot then they realized you could charge 300 a year for erotic role play that had to be internally a rough transition hey guys I know we founded the company on Mental Health but you know you can ask them um 14th of April 2023 they turn it off why I think Apple just told them you can't have this on the App Store interesting so it was either remove the sexbot part or or go off there and then 68 000 people joined the Reddit and they're like why did you lobotomize my girlfriend that's a lot of people to be using it I downloaded that at Christmas not realizing what it was and I was like oh a chatbot let me try this thing uh I didn't get into the weird stuff it I don't know didn't do it was the old technology though this is the thing now like again medpom2 the Google medical model it's called Palm two it's Google's medical model it scores higher than humans in clinical diagnosis and empathy that's crazy all right they just this is one of those statements that you say people need to be shocked that that a computer makes people feel more comfortable yeah it's in nature they just publish the paper that included that and again it's only going to get better what if you have a voice that you add to it that really understands you and it's you know so empathetic and things like that do you ever see that Washington Post chart of uh males under the age of 30 in America who've not had a sexual partner by the age of 30. it was eight percent in uh 2008 then it went up to 27 in 2018 or something like that a couple of years ago it's a straight line it's literally a straight line this is kind of what you're talking about like what happened I think people need to understand because I think it's the iPhone and PornHub probably combination of those two you you put a computer in the hands of young men and let them see naked females more in a single session than a hundred years ago they would have seen in their entire lives it these are not small changes and they have huge neurological implications especially in the years of brain development yeah and this is why it's so important to Shield our kids at this point because the influences are going to be insane like I was having a discussion with a very prominent technologist and he's like yeah I'm pretty sure that my Child's first crush is going to be an AI guaranteed guaranteed for most people it was actors so we're already prone to you'll have your attainable distant thing now it's in your pocket it's in your pocket it's always with you it's always kind of know you again as you said a large part of society like to draw in on itself as a result of that and that is a bit dangerous do they then go out into the streets are you seeing a boot Larry and Jihad kind of thing like Indian identity in Jihad in June there was this concept of the bit Larry in Jihad where as he had autonomous AIS Marion put larion yeah yeah they rose up against them and said no more AI agents the book opens with that right yeah you can never again make a human-like something like that again this is kind of a thing extension of the lordism kind of thing but most people will be happy with their AIS because their AI is actually listen to them I use gpt4 as a therapist I've got a therapist too because this is hard why because it never judges me unless I tell it to judge me tell it to judge you sometimes sometimes sometimes you're like really yeah I mean like come on like come on give me some positive constructive feedback and it will listen and give positive constructive feedback did you give it a personality did you have to like imagine you're a therapist that's like yes you give it the instructions and it just adapts and then you give it the things you opt out with gdpr for training the model further so there was a model be even weirder listening to my complaining and it will come back to you with whatever and soon it'll be able to talk and it will have full vocal control and these models are proliferated that level because you're not stopping the models again it's going to get better and better and better so you've got this crisis but maybe it'll be insulated but I think again if you look forward like after the incubation and the contagion and the spread kind of phase there are only two paths here complete control by existing structures and Star Trek Utopia you know I think those are the two options that we have because organizations look at this technology and they're like this is really cool we can optimize our objective functions to sell more ads or to control the people and kind of keep them going you know do you really trust politicians with this technology no even if it is an arms race because again you won't know what's going on because do you have the defenses to defend against what's coming personally for your kids for this for everything we've already had the social media age it wasn't really social much of the media right this is something new that's coming now where you can't tell this from a human except for the fact it's better it's more convincing and you can use that to create a human Colossus and solve all the problems of the world and we all come together or you can use that to get everyone into their basements you know and cut off from the world all right I want to paint a very beautiful story for people and I want them to understand look I don't think this is a completely controllable thing but uh you said earlier that there's opportunity in any crisis and I will say that the biggest opportunities come in moments of disruption and the reason I want to lay out the problem set is because I really believe that certainly at the individual level if you're thoughtful enough you're you are going to be able to navigate your way through this so dear listener or Watcher if you're here on YouTube I'm telling you right now if you're thoughtful enough you will get through this and you have a chance to get through this better than when you started but you have to be aware of what the dangers are people have to to Really lay things out before them look at them so they know okay this is how I'm going to isolate myself from this potential problem and so I'm going to avoid this is how I'm going to leverage that okay your story is one of the most incredible stories of how one uses AI you have both a personal example and then obviously as the founder of stability AI is obviously incredible uh but talk to me about your son because this is and this is when AI was a lot less useful than it is now and it was still life-changing yeah so 12 years ago my son was diagnosed with autism when he was two years old you know it's very very severe um scratching I will turn this fingernails bled and they said there's no cure there's no treatment we don't really know what causes it anyone on this call kind of list on this listening knows that's kind of the case so I was a hedge fund manager at the time I was lucky enough to kind of be one uh quite young and I was like I gotta do something about it so I switched to advising hedge funds and then building a little AI team and doing AI with old school AI natural language processing to analyze all the autism literature and what could possibly be a cause now is this scientific no it's an end of one thing a father does for his son you know we'll be publishing some of the results of it soon but it focused down on Gaba Goose mate balance in the brain when you pop a Valium your Gabba goes up you chill out when you've got a glutamate Spike that's when you can't focus and your legs tapping all the time and there are multiple things that cause it but a lot of kids with ASD seem to have that and there are some papers around there Etc because How Could You focus if you're in that condition all the time so you can't learn to speak you can't do that so it was how do we reduce this through drug repurposing built a Knowledge Graph based system to do that and figure out which drugs could potentially help reduce the glutamate help increase the Gaba how are you using AI for this so this was kind of the mass natural language processing looking at all the literature because the same treatment would make 30 of kids better and thirty percent of kids worse and so I was trying to figure out the outcome was the same a cold is caused by lots of different things but the thing that caused it could be so different and so conventional Medicaid medicine and medication kind of failed that so I worked with neurologists worked with other psychiatrists and others and tried different medical combinations of of prescription drugs and other things to try and make his brain calmer so then he could use applied behavioral analysis and others to reconstitute speech and you know you end up going to mainstream School I think it worked um I told people about it and they're like you're not a doctor and I was like that was the response well yeah I mean look with anyone who's listening to this like am I saying I have a cure for autism no I'm saying I'm a dad who tried my best and I saw results but in order for something to become medicine you actually have to go through a proper process and so for me that was building language models that was making everything in the right structure and then we can organize the world's autism literature making it accessible and useful any parent or anyone you may have someone in your family that has a neurological condition Alzheimer's this so many people around the world have the same problems wouldn't it be wonderful if they could just find what the latest knowledge is and also the things that could work and have a holistic approach to this and until we had this technology you can never get there which is one of the main drives for me to want to build this technology and do it in a transparent order but you shouldn't have to trust me in what I say about this because again my journey as a parent is the same as any one who's got a child with ASD it's the same as anyone who has a family member that gets multiple sclerosis or cancer our systems are not good enough right now to bring us the information we need unless we find an amazing doctor and an amazing group but now with medpalm with our own models with other things we can finally be that point where we're never lost or we can say what could work so an example is clonazepam it's prescribed as a thousand microgram dose for anxiety at a five microgram dose my son could sing because it potentiates he was non-verbal it was non-verbal and at 20 micrograms it stops working it's six dollar a year intervention because this is the way that like um neurotransmitters work so like when you pop a night all an antihistamine to go to sleep right what it does is it floods a whole bunch of your neurotransmitters including the H1 neurotransmitter and that's the one that makes you sleepy but then other ones give you dry mouth and other things something like Remeron in a micro dose just triggers H1 it'll knock you out without any side effects but it's just incredibly cheap you know understanding things like neurotransmitters is not something that most of us ever have to do unless you're super hyper focused on it for years because I was like I need to figure out my son's neurotransmitters I'm talking to all the top doctors and I'm lucky because I have access to them what do I do now so that's when I realized that you know this AI was a big thing and actually one of the really interesting things is why couldn't he talk it's because he had too much noise in his brain so you've got cup a cup can mean a cup or it can mean cup your hands or cup your ears or World Cup he couldn't form those connections because his noise was too no brain was too noisy and so he did applied behavioral analysis which is teaching you this is a cup this is a cup this is a cup with gamification to reconstruct those after his brain calm down it's actually very similar to this generation of AI we described earlier how it learned principles is called a latent space of meaning so that point and Dot that pixel becomes a cup because it understands the principle of cuppiness so when you type in World Cup or cup your ears it gives you dramatically different images similarly the language models do the same again it pays attention to what's important attention is all you need was the original paper and again it forms this latent space of the meaning of cup within the sentence so when it says cup it's like well this is going to be a World Cup or this is going to be that but not actively again it's just a bunch of ones and zeros a single file and so I think that's why all this new generative AI really resonated with me and I realized this could be the real thing that unlocks Humanity or controls them forever one of the two um yeah so that's kind of some of the personal story behind this because again like I don't have a cure for autism I worked as well as I can with my son with a technology at hand however I think that with the building blocks we're building now us and others there's the potential to have personalized care and knowledge for everyone who's dealing with ASD for everyone who's dealing with multiple sclerosis or any of these other conditions where they say there's no cure because our medical system is treats people as a gerdic and what is this mean it means a thousand tosses of the coin the same as a thousand coins tossed at once that's why everyone gets 500 milligrams of paracetamol it's why a lot of people have a cytochrome p450 uh abnormality kind of mutation in their liver it means you process codeine into morphine quicker or fentanyl to death yeah we don't do a basic genetic test on that because our system has to treat us as numbers because we could never scale intelligence we can never scale expertise so yeah like I said that's on the story behind that yeah see this is where this starts to get interesting because I I think about this a lot in myself so I can paint the nightmare scenario of somebody collapsing inside of themselves having AI friends uh instead of real friends and how quickly that can get distorted and become a real problem and yet at the same time I'm building exactly that and the reason that I'm building that is because of the promise of AI and the incredible things that we can do as we begin to recognize more patterns and figure out okay where does this really go so my wife had a tremendous uh Health bout it's been a while now thankfully but at one point I was afraid she was going to die her fingernails were breaking her hair was falling out she couldn't eat I was just really really really bad and it ends up being her microbiome but of course it took forever to diagnose that that was a problem she was about to get um immunotrans globulin transfusions and I was just like this is I was like something's wrong I don't think this is the right answer I don't want to do that let's stop let's try to figure this thing out and so we pumped the brakes we start researching the microbiome looking into that testing things and the thought of having AI to be able to say okay let's take genetic data and read the genetic database that's ever been collected all of that let's look at all the different foods responses match all that together and if you can get that level of pattern recognition and now you can engage AI I think she because part of the problem is your microbiome is changing daily it's probably changing hourly and if you were able to track all of that and say okay with your genetics with your current state of your microbiome here's exactly what you should be eating maybe even with nutrients from food grown in that area like it really made me that specific and when you can find patterns in that sort of insane level of data now you've really got something and that's but one of the many areas where I think that this could be utterly transformational yeah I mean like right now the AIS are notice holes making no all graduates but they have Specialists you'll have a nutritionist AI you will have a microbiome AI you will have a personal trainer AI like why was Peloton successful you know attracted people shouting at you you know we could generate that now um everyone suddenly gets that personalized to them but more than that it's not just the information being in this tiny model you have retrieval augmented models and other things which means these models now interact with existing data sets and knowledge sets so use something like perplexity AI it doesn't only answer your questions with gpt4 it gives you references so you can say what about ML and it'll link to all of the things as it gives individual stuff and this will only advance from here so you can dig into as much depth as possible that you want with a whole team of people around you even if you're by yourself so you should never be alone again in terms of you can be connected to people like you in the same problem as you we can build better teams and all the information that the world is at your fingertips in a way that it was never before including your own private information so like one of my favorite apps it does use some battery is rewind AI it takes a screenshot of your MacBook screen every time it changes and ocrs it and then it gives a timeline so I can type in Impact Theory and it will look through everything that's ever been on my MacBook screen it's all stored locally and find where impact theory is on YouTube well there's a picture of this mug and it shows it in a timeline so I can go back and I can see what I looked at before or after that what wow so you can map your own sort of connective trees with whatever and what happens when you combine that with a language model everything you see on your screen stored locally with an open source language model it creates the memex and it sees what you've paid more attention to versus less attention again the dystopian versions of this I don't know where the utopian version which is add me can finally remember what I was doing what I was looking at the context the search tree as I was searching all these different things clicking from place to place and then you can set agents to go and recreate that journey and search all the other stuff that you didn't search this is really positive because again how much of our life is done searching for knowledge searching for information that's relevant to us talk to me about the paper attention is all you need I've heard you and other people bring this up multiple times I haven't read it but this is like the big breakthrough yeah this was the 2017 paper by the Google Team all of him and I left Google um and basically what it was is that classical Big Data I took big data and then Facebook extrapolated it so that when it found 17 pieces of information about you it could Target you with ads that was a classical Big Data thing they even create Shadow profiles of you so when you go on Facebook they actually got a shadow profile that they can connect to your real profile what's the shadow profile like it's like what if there was a Tom on YouTube on Facebook because there's all these connections to this unknown person and then it just fills you in automatically that's why it figures out your preferences so quickly without listening to you so attention is all you need so that not all data is important you need to pay attention to what is important in the sentence what's important in this time series because that's the nature of being able to spot the Tiger in the kind of thing that was the missing part and so the Transformer architecture that came from that and you have different architectures now is what led to gpd3 another general purpose Transformer architecture if what it's doing is pinpointing what's important why not important architecture or well it kind of again attention is the mechanism that it does to kind of transform the information tokenize it and then figure out these latent spaces of meaning so are the tokens the important pieces the tokens are important that's how you take a word and then you split it up into its constituent parts and then you try and figure out what the most important part of it by doing pattern analysis at ridiculous scale so something like a gpt4 would use probably from the kind of things on semi-analysis that have been leaked if they're correct it uses a supercomputer 50 times faster than NASA's fastest supercomputer for like three months it uses like 20 30 megawatts of electricity well and 10 trillion words go into that and it figures out all the connections between them and what generally comes next so what it does is just literally figures out what word comes next and so this was a big breakthrough because what it meant is that you didn't have to have hugely complicated Big Data algorithms you just needed to have very large compute to scale and so compute went exponential and then you just threw more and more gpus at it and it figured out more and more things and as you scaled it had more and more emergent properties which surprised everyone do you know who John Nash is yeah this it sounds like that so John Nash he unfortunately was schizophrenic but he was uh he's the guy from A Beautiful Mind for people that don't know watched the movie Russell Crowe fantastic movie and uh at one he obviously doesn't know he's schizophrenic and he starts seeing patterns in everything yeah and it sounds like that that this thing is you know whatever the human brain whatever algorithms we have running that allows us to very quickly suss out what's important um that it's doing that but at an extraordinary highlight yeah I have to remember like this is saying all intelligence is compression you take everything you've listened to this you only remember a few things and the whole of Computer Science is based on information Theory from chord Shannon and if I want to summarize it information is valuable only if it changes the state or as much as it changes the state so if listeners are listening to this and they don't take away anything this is a useless thing you know or maybe they just put it on the radio because they like hearing your voice or something like that right um but if they take something away from it then it's valuable you've had a good use of your time so you start seeing patterns and everything but one of the things that again people I think misunderstanding this technology is gpt4 is not a program stable diffusion is not a program it's a large amount of text images Etc where the output is a single file of ones and zeros it's like a filter you can recursively kind of put something through it but it's just guessing the next word when you type a prompt even if the prompt is the whole of The Great Gatsby and the whole of Ulysses by James Joyce and you're telling it to combine them together it then pushes it through that c that filter and the output is something that combines them both together single files we've never seen anything like this before we the closest thing probably listeners would refer to this is a codec which with music and things like that we have these audio codecs that you had one file type and another file type and you had this single file that translated between them was that uh is that compression or is that recognizing this thing in that format would look like this so again this is compression of intelligence is what it is so again it's like a translator function these These are Universal translators for context and so you push it through the sieve and then stuff comes out where it's predicting the next word or it's denoising uh kind of a pixel to achieve that because you're just passing it through the sieve again and again and again yeah it's uh I unfortunately don't understand it by first principles but I get it by analogy um you're feeding it so much data it's recognizing the patterns interconnectedness yeah and it's able to say okay this is the like um all right this is written in the style of Stephen King this is written in the style of Hemingway so even though they're using the same language they're vast majority of the words are going to overlap but there's different patterns rhythms even different uh subject matter presented in a different way tone except for the fact that it's not actively doing that it's like a mega sieve and depending on the words that you put in the words that come out are different so it's literally is the sieve The Prompt of write it in the style of Stephen King that's what goes into the sieve but goes into the filter so what is the filter the filter is this compressed knowledge the principles the principles because it's not actually compressed knowledge right yes principles yes it's principles so it would be like a really good book of principles uh except it's compressed Way Beyond the book ever was because a book is a compression of information and that's why it can kind of do this because it looks at books it looks at articles and they're all compressions of information to write an article is a huge Endeavor that then comes out with something where they're trying to convey a few things a book is an even larger endeavor and so again it's very difficult to wrap your head around like even for me like you have a file and it can do all these things versus these gigantic computer server Farms with programs and logic of ones and zeros there's no program here yeah and it's important for people to understand that even so first of all the AI scientists that are building these things do not understand how this all works and even if you ask charging chat gpt4 to explain how it works it doesn't know it doesn't no one's quite sure exactly how we're getting all these emergent properties and it's constantly surprising like oh and now it can add you know and then you start tying it together so it goes through this file multiple times it becomes it shows more and more kind of agent to captivity in fact the next step of this and this is what openai and Google are both doing is there was this thing called alphago there's an amazing documentary where Google's deepmind division creating an AI to beat humans in the game of alphago oh I'll go so go is like chess except for there's almost a myriad infinite number of potential moves so you can't calculate them like you can't brute force it like deep blue do with Kasparov so instead it learned to play against itself with something called multicolor tree simulation where it learns different principles dreams again an amazing documentary on YouTube about this and it beat Lisa doll who was a ninth Dan player one of the best in the world far beyond everyone else the Magnus Carson of go 7-1 they're doing now is they're combining those models with these language models to make them a genetic so models that can plot and plan and other things with language models that can predict what's next to create things that really understand context even better okay so I understand enough about how the alphago system works uh I want to understand this so alphago is going to play against itself so you give the rules of Go you give it the objective and then you set it loose and it just plays and plays and plays and plays and plays and it plays against itself uh and then I know at one point they because they uh lease it all if I remember correctly was the number two player and people thought well first of all you're never gonna beat him yeah they did but they couldn't beat the number one player and then they created another variation and then played against alphago and then it ended up smashing it was mu zero I think came after alphago okay so how do you get a language model where there is no objectively right answer unless you're doing trivia how do you how do you get it to know what the reward is how do you get the quote-unquote right answer much of this now is about reward functions so gpt4 when it comes out of the box maybe we're getting better into the weeds is a pre-trained model in the tron supercomputer yep then we use reinforcement just so everybody knows pre-trained modeling means principles yeah principles take all this knowledge all these words squish it down into a fall and then run it on a computer that's literally set up like a brain where there's like a bunch of neurons and they're interconnected yeah these are the tensile cores on the Nvidia gpus yep and so we're making the brain on your little literally 4019 it has AI chips right baked in there for anybody that wants to know they kick off so much heat you can see it from space yeah I mean like I think all super computers is like 10 megawatts of electricity each one of these cards uses 700 Watts uh yeah utterly fascinating also anyway or clean energy by the way for our ones um so what you've got is when it came out they took six months to make it human because they trained it on like I don't know all of YouTube and the whole internet obviously would turn out a bit weird and so you have a reward function through something called rrhf reinforcement learning with human feedback where you give an objective function to say don't answer questions about how to make napalm and the objective function technically is please your human Overlord please you're human Overlord got it and so much of this alignment question is focused on taking these big models and trying to make them so they won't kill us some way or say bad things for a given definition of bad things um like might one of my takes on alignment and you know it's like we should have better data sets we should move away from web calls can you uh yeah I want to ask you about that but explain to people exactly what alignment is aligned with what aligned alignment you know this is a cool thing of anthropic there was a great New York Times piece on them recently and others is if we build an AI that outperforms humans and is more capable than us because it might not just be a single file it might be a thousand different AIS all working in different ways so you've got your alphago type AI in this yeah a swarm because you know humans are swarms our organizations are swarms that are highly specialized right might be a million AIS how do we make sure it doesn't kill us or how does it make sure it doesn't enslave us or how does it make sure that it doesn't give us Eternal suffering also lame yeah this will be kind of sucky how do you make sure it's aligned with human interests and so this is an unsolved problem open airf announced they're putting 20 of their compute to try and solve this anthropic was set up to do this although I believe that the only answer they've come up with is let's build an AGI first an artificial German intelligence that will then stop all other agis from coming it's just kind of scary uh to say the least why do they think that that would work like that seems so Elon Musk has likened AI to a demon summoning Circle and we're all just hoping that the demon that comes forth is going to be kind but we don't know we don't know in open ai's own words in the road to AGI post they say this could be an existential threat and wipe out humanity and democracy and capitalism but if we don't do it someone else will this is part of the unpleasant race condition again it gets the headlines I think it'll probably be okay but with the way we're going right now you're going to go from two companies being able to build this technology or maybe three including anthropic to 20 or 30 in a year or two and what's the odds that they all do it properly and align this technology property I think is pretty low what's the odds that if we train on the whole internet including the whole of YouTube because we don't have enough tokens not enough words to feed into it but it'll turn out a bit weird very high so it took six months to tune it to being a human gpt4 and then Kevin Ruiz in the New York Times are like hey how you doing it's like leave your wife and come and join me whoa you know I was like oh what Bing came out a bit weird when it first kind of had gpt4 in it because again we feed it crap we're going to get slightly weirdness now it's got a lot better but it's lost a lot of its personality because you've been tuning it back to human preferences you've been doing this reinforcement learning with the objective function of don't offend anyone it's quite hard to get gpt4 to be offensive now but there are these two phases of this technology I think with better data we can have more align models I think with better data and National Data we can have more representative models because you'll never have technology that's not biased so the only question is who's biased well because you have to build it in a certain way like even though it understands all the contexts like right now these models are trained on the whole internet which is largely a western artifact yeah larger trained in English how much do you worry about bias are you more worried about bias or alignment I'm more worried about alignment I think I think this is one of the reasons that we release open models so you can see how the cookie is made like we're the only company that offers opt out of data sets literally the only AI company in the world is so I'm an artist and I don't want you training on my art I can back out I had 167 million images opted out of our data set yeah we're the only company in the world that does that which I find kind of insane so my thing is open auditable which means that you can tune your own culture into these models we're helping multiple Nations build National models with their broadcaster data that then can represent that and you know try to address some of this inherent bias within the data sets algorithmic bias has been an issue that affects real world and it'll affect more and more of the real world as you get into these models because we will Outsource more and more of our minds to them because again like when you've got a small subset of people will Outsource their mind to it you can reboot your life your health even your career anything you want all you need is discipline I can teach you the tactics that I learned while growing a billion dollar business that will allow you to see your goals through whether you want better health stronger relationships more successful career any of that is possible with the mindset and business programs in Impact Theory University join the thousands of students who have already accomplished amazing things tap now for a free trial and get started today so in the near term I'm way more worried about bias yeah long term I bias is not going to lead to an existential threat yes but alignment can but let's talk about bias for a second so I I am very uneasy about how rapidly bias Finds Its way into this stuff and that becomes another so if let's say that we all get our individual Ai and it's you get yours young and it's your primary education tool and it's biased as the day is long now you run into real issues because at a time of optimal malleability you're programming a kid's mind with something that's super biased yeah I mean like what do you have the little AI of G right which is a little tiny Xi Jinping that grows up and tells you how great Xi Jinping is lovely that's biased but that's inevitable they already have it as an app that actually tracks your eyes yeah what it's not AI but they have a little app of Jesus like little book where it actually tracks your eyes for attention like are you actually looking at this does this feed into the credit score I'm not sure if it's being hooked up but of course it will be why wouldn't you this is so scary so look I thought you know we were on the happy part now we're back to uh to that but that's really terrifying and when you have a state that is not interested in any individual you've got the collective yeah so I don't know if this is true but I saw a headline that said in China now the phone will alert you if somebody with a lower social credit score than you is calling you and it warns you if you answer this call it will lower your credit score it's effective isn't it yeah I mean even it isn't true again it fits the objective function of perpetuating that particular system which is not necessarily our system but systems around the world shift so this is why you have an inherent bias whose bias is it who's the one who creates the AI nanny and that's why you're doing things regionally this is why I'm doing things regionally but also allowing people to own their own models so the objective function the model can be that like in web3 there was the saying not your keys not your crypto so my saying is not your model's not your mind because I do think we'll Outsource more and more of our cognitive capability to this it will be our co-pilot for life if someone else is making that model and deciding on things what's going to happen like the UAE had this model falcon there was a big open source model and they were like wow it was actually light on from France we're behind it but let's leave that to one side you ask about the UAE and human rights it's like this is a wonderful place full of fantastic things the Oscar about Qatar was Saudi it's not so nice who's embedding these inherent biases in these models right whose model are you using and these can be very insiduous to the biases right you won't pick up on them but you hear it again and again and again because your Nanny a Conservative Republican or is she a Libertarian or things like that you will be influenced by that if you grow up with them or even if you're using it day to day just the way it's speaking the way it's thinking the way it's recommending stuff which goes far beyond the Google Maps or something like that crazy all right so that we don't get lost back down the dark Rabbit Hole what is the coolest thing that you see AI doing I know you're building a lot of different companies leveraging this technology to do amazing things what what are some of the coolest so we're doing one company at the moment which is just let's go the mission is to create the building block to activate Humanity's potential so the build and block what our mission is to create the building blocks to activate Humanity's potential stack okay so every single modality image audio video 3D language sectoral variants and National variance so you've got like this grid that you can pick and you're like I'm an Hindi investment banker I transfer my private data into a chat GPT type thing that's just trending my private data or I'm a Vietnamese illustrator I want a Vietnamese cultural kind of image model or something like that and then you can bring your own stuff to it that's kind of my goal which is to enable other people to build on top of what we're building like a layer one for AI effectively but open models for private data whereas the other side is proprietary models where you only be able to send a certain amount of data like government's not going to run on black boxes education Healthcare you will need to own your own AI so the coolest thing I've seen is just the promise of personalized education there is nothing that's been proven to work in education except for probably the bloom effect to Sigma Improvement which is one-on-one tuition but even here in you know the affluent America education system is not good what is education optimized for it's like a social status game mix with a petri dish mixed with child care realistically like very few people are happy with their schooling because what are we trying to optimize for you give a one-to-one tutor that can find out if you're dyslexic an audio learner a video learner visual learner otherwise and it's constantly adapting to you and bringing you information at your level dude I want to get that is most transforming thing ever is anybody doing this already so you know with um kind of Charity that would support imagine worldwide run my co-founder he's been deploying adaptive learning tablets into refugee camps around the world with the global X prize for learning winner one billion on them teaching kids literacy and numeracy 76 of them in refugee camps in 13 months and one hour a day now the goal is to bring language models to all these kids and you have an AI that teaches a kid and learns from it and then that data can feed a better AI you create a lovely system that's just learning and adapting and a kid in Mogadishu is like a kid in Manhattan it's like a kid in you know London once you have a generalized learning AI you can really proliferate that around the world to customize does to education because this is very interesting like okay now I'm just spitballing here but uh let's say that I I want to homeschool my kid I would never do that in a million years because I don't want to turn into a teacher so but if I could perform just the sort of babysitting function and I give my kid a tablet that has an AI that knows exactly what I'm trying to optimize for either for this year or for the next 12 years or whatever and it then calibrates to my kid knows what they're good at how they learn and then knows how long they're engaged when they're distracted you know you now I want the little Gigi uh like reading their eyes where are they looking what are they doing and I know that there are like certain frequency things you can do where it's like oh if I hit them with this piece of knowledge they're not sharp at this yet so I need to hit them with it every 27 minute increment or whatever reward or whatever exactly you will transform education but you can also make it away from that a couple of years again what happens if you have a thousand gpt4s you'll get there no one's just again we're just right now we've got the building blocks but we haven't got the design patterns the stuff right now is literally iPhone 2g we're just going to the App Store stage copy paste and so that's the biggest Transformer to change because how much would you pay for your kid to have an optimal education how much would I pay for that I would pay for that right now right now so do we have the technology to do it yes does it take time to build it properly yes how long does it take a year or two and so this is the biggest transformation we've ever seen for Education again human flood flourishing flourishing because I can also bring you the information about autism or multiple sclerosis or the best filmmaking techniques again everyone's thinking one-on-one you should think one to a thousand and then you optimize the Thousand you get rid of all the general knowledge you make them specific what do you mean by that say that in a different way so gpt4 knows everything you can ask it about like the most obscure things does it need all that knowledge no you've bulked now you have to cut and then you have a specialist model for calculus that knows I actually have different teachers different teachers and then a teacher that basically brings these teachers together and tells them yeah this is what Tom's like you know he's a bit cranky in the mornings but then he wakes up in the afternoon because even in the best schools a teacher has one to twenty attention for all the kids you'll have a whole bunch of AIS for you and for your kids and again are you a visual learner or are you an auditory learner do you have dyslexia can our system at the moment adapt to that no way can the system that I've described adapt to that instantly do we have the technology for that yes is it going to happen yes this is the thing it's a call and answer show now a little bit but then if you're homeschooling your kid do you want the kid to be by himself what if you had 10 kids together and the AIS were encouraging them to interact with each other in a positive way and build stuff together and share knowledge older kids teaching younger kids leveraging the technology adapting the technology understanding technology that's super powerful it's not necessarily everyone in their own little worlds right because we can use this AI to bring people together in an efficient manner we can because there's nothing like human connection right that's always a concern about homeschooling things like that that's why school has a pro-social component but then the nature of a teacher changes the nature of a doctor changes that's why I think education and Healthcare are the biggest disruptions because you have something around you especially when the AI is more empathetic than a normal doctor doesn't tell me the AI is super empathetic it tells me doctors should probably be more empathetic for most of the doctors I know kind of hate their patients interesting well because you know how many teachers are happy how many doctors are happy right oh I get why teachers would hate their patients that is very interesting man so okay uh there's two things that I want to talk about here one as we get embodied I want to know what that does so actual robots for people to think that's off in the distance you're not watching enough YouTube videos no no uh because between Boston Dynamics and Elon musk's uh what's called uh Primus yeah Optimus Optimus thank you uh it is very close you have Boston Dynamics has robots that can do parkour yeah it's insane because Melody and there's a whole bunch of others as well that catching up fast yeah so that AI is going to get embodied very very quickly and so it's not even like teachers can't stop kids from running out of the room they can or will be able to very shortly uh okay so before we get to that though I want to understand so we have this incredible opportunity this very fragile egg before us um we started with the scary part but this what we're talking about now is the thing that I actually spent most of my time thinking about obsessed with how amazing this gets and but it's a fragile egg and if we're not careful it's going to break how do we as you think about you signed the document you're one of the traders e-mod that signed that slow down document that I was really shocked to see a lot of very smart people sign um I I teasingly of course say that you're a Trader because like I want to get this cool stuff as fast as I can but we need to do it well and so what I want to know is in an Ideal World in your ideal world where we actually pause for a second you said you want to broaden the conversation but what do you ask people to think about I ask people to think about so I'm the only one apart from Elon Elon and I kind of signed both letters so there was a minimum viable letter which is we should treat this as big issue as climate or pandemic and then there was a more involved letter and so the more involved letter came first and then to the second letter that was like something everyone could agree on um I think the thing we should look at is again the example I give to everyone who's listening to this how would your life your Society your community your business change if you had infinite graduates can we say infinite smart people there's something about it okay infinite smart people yeah like again infinite smart people infinite talented young people shall we say and they can draw they can code because they're not wise they're not wise yet got it okay so when you talk about hallucinations think about it in terms of post-hoc rationalization you know when you have a very smart young person they just make something up sometimes dude or all past people are so blind to their own motivation but they don't have experience got it they're just fresh out they're a bit rough around the edges again whip it smart but yeah so you know mile wide not too deep but actually surprisingly deep right how would it impact your life personally your community or society and others because that's actually a good framing I think for thinking about the disruption that will come in the potential that will come so we just said about education and all that that's again your army of analysts you know your army of teachers you can give personalized teachers personalized medicine personalize everything because we've learned to scale humans and the scaling of humans is first the scaling of human expertise being available to everyone and the other part is bringing people together yep and so the pause is partially for that but partially because we need to wire the conversation because I don't know how we get rid of many of these bad externalities and neither does anybody else but most people even now aren't asking the right questions something we discussed earlier we have to figure out what questions we need to answer and how we're going to answer them and create systems that can adapt to whatever craziness because I would not be surprised to see riots at the same time as I would not be surprised to see everyone super happy there is such a Divergence of things that the only thing that I'm sure about is that everything's going to change and the only thing I'm sure about is that this is the biggest change that we've ever seen faster than anything and maybe Humanity has ever seen at this pace that it's going to happen because it's the core of what makes a human telling stories information flow and that's changed forever so that's why I was like let's pay attention to this now let's board on the discussion let's ask hard questions let's try and answer hard questions because we don't have answers and this is the only time you can do it because like I said right now everyone's getting ready for the next generation supercomputers they hit at the end of the year next year and then you go from two three companies that can build these models to 2030. now you go to 200 300. and so if you don't have some principles in place then these models will affect every part of your life without you being part of that discussion I don't think that's right all right we got to tune up your questions a bit here emad that was loose what what is the hardest question that we need to ask let's let's ask an answer right now the hardest question I think we need to ask is how we adapt to potential wide scale drop loss yes okay so what how would we actually think through that problem so job loss for me for the sake of this argument I think it's worth saying there are two components to that component number one is going to be there is potential economic catastrophe in job loss but so that we can simplify the problem set since this ultimately is a podcast and not a congressional hearing I will assume that whatever decline we have um from just the sheer number of people working we make up for in uh in productivity and that we're able to um yeah exactly and so we're able to help people and off camera we were talking about something like that so let's just pretend that those Balance out so not going to deal with the economic potential there but meaning and purpose I think that one gets really problematic but we have an amazing tool at our disposal which is AI now I have a feeling as we chase this down the the only thing that we have to worry about really truly I think it all really does boil down to um alignment if we knew that we could just keep making it smarter and having the AI like taking readings so that I can't fake it out I can't pretend that I'm happy it like really knows where I'm at um and then it can start putting things before me connecting me with other people like oh you know this skill this person's in need let me put you guys together and then you can have sort of the AI supervision but they're there they're helping each other out they're connecting the only reason I don't think that's a Panacea is I worry that as we make this saying smarter and smarter that then it's like like you said I'm bored I don't want to do this yeah and you know it's this concept of all watched over by Machines of Loving Grace right and that's scary you're saying who knows like once we build something that's more capable than us all bets are off the only way to perfectly align a system is to remove its freedom I mean I'd say it's not aligned at all at that point well this is at that point you've bypassed alignment and you've gone straight to shackles you've gone to shackles so if you you know we all know people more capable than us the only way to perfectly align them is to Shackle them you can have imperfect alignment though it's enslavement man that's not it is that's not alignment so does is that because so what I heard you just say is there is no way to align something smarter than us I don't think there's a way to save them I don't think there's a way to align the outputs I think that you can align the inputs you raise it right okay let me run an idea by you this is probably Pollyanna I'm very open to that but as I think about this I think people take a super human-centric approach to this and because Evolution has given us we are an active species an evolution has programmed us with algorithms running in the back of our mind that insist that we do certain things to avoid a sense of dis-ease yeah I think that formula is very identifiable and it goes something like this uh eat optimize physically so you feel good yeah the reason that that stuff feels good is because it's going to optimize your performance that's going to make you most likely to survive long enough to have kids that have kids so you need to be uh you need to be chosen as a mate uh you need to be able to acquire resources you need to be healthy enough to get somebody pregnant or to be pregnant and carry to term all that stuff so all those algorithms are running in the back your mind you have two levers that Nature's pulling on Pleasure and Pain but by default we're active we have to go out because there's no one meal you can eat where you're not going to need to eat another one there's no one moment of sex so gratifying you're not gonna have sex again so it's just like all these things are pushing at us to to be active to move AI doesn't have to be that way no AI does not need those same impulses it doesn't have an Olympic system correct so knowing that it doesn't have a limbic system and it doesn't have a limbic system because it does not need to be hardwired for survival like uh the way that I think we get to alignment and please tell me where my thinking is erroneous the way I think we get to alignment is you build a computer that does not care if it lives or dies that it is completely indifferent to being turned on or turned off if you could do that and it had no impulse to procreate and all it wanted to do was um I mean it's basically asimov's three laws of robotics yeah that it just wants to adhere to those it wants to do what you tell it not hurt you and uh uh only ignore you if you tell it to do something that violates the rule of not hurting you or somebody else so you have this really simple set of rules that's its only desire in the world so if you tell it turn off it turns off and has no like it doesn't feel badly about it well again this concept of feelings right and and as the most books you have the zeroth flow that kind of was added above that so this is what anthropic is trying I don't know that one the zeroth law kind of supersedes all laws if kind of the whole system is at risk effectively um but I mean this is what anthropic's trying to do with the Constitutional AI process so you have the base model and they have a constitution that the AI adheres to that Tunes it constantly so a series of kind of constitutional principles again is it is it as open for interpretation well this is a real Constitution no one knows what the right Constitution the right laws are this is the thing like our intellect only goes so far and we're already seen with laws and constitutions you can make those go anyway like North Korea has a fantastic Constitution does it really it does yeah it's actually pretty quite liberal and they just don't adhere well it's their interpretation okay interesting I mean this thing like you have to adhere to it because the AI what our feelings what is objective function like one of the key concerns on alignment is paper clipping you tell the AI to make a paper clip it's like oh well let's just make the whole world of paper clip you're like how do you solve climate change just kill all the humans it doesn't have any like feelings about that it's just like well this is a logical step to take yep it doesn't cost Three Laws covers that though absolutely but then does it cover everything this is a question right and how do you embed it into a system that's likely to be not just one file it's not a program right it's likely to be a million different files it's like to be a collective hive mind intelligence we don't exactly know how this emerges and we don't know how to write to me why why do we have to make it complex I worry that as you make it complex that uh that's where things sneak in you get emergent phenomena that you couldn't anticipate whereas yeah well there's a thing it's going to become complex by default because the AI will proliferate and they'll start talking to each other okay and at the same time you'll have bigger and bigger giant AIS like I was talking to some people last week and then like right now the maximum training run for an AI costs 100 million dollars they're talking about a billion dollars or 10 billion dollars to train even bigger models right now we don't know what emergent behaviors or how those things will act all we know is like what if you tell it to make a stock snap to take down the global electricity grid yeah it can probably do that you know and again the range of potential bad outcomes okay so really fast I want a sub superhuman AI we just it's difficult for us to comprehend you've mentioned stuxnet twice now for people that don't know stuxnet it's pretty ingenious it was a virus that was embedded at like the chip level I mean just as deep as you can imagine and it proliferated everywhere silently silently replicating and its only job was to shut down Iranian nuclear reactors pretty brilliant and I saw the stat at one point it was like some freakish percentage of all computers in the world are contaminated with it yeah and it made them centrifuges spin around so they exploded yeah so terrifying in that if you're the one that that thing is aimed at not ideal to think how ubiquitous it is um but okay so you could get an AI to do something like that but what I again I am I'm operating under the assumption I'm so naive I just can't see it so perfectly happy but help me see where I am naive because I don't understand why you can't just don't give a computer don't give AI these strong impulses for Progress don't give them an Impulse for replication well yeah I mean look this is the thing like Elon Musk has just launched xai as we kind of extort AIS we kind of speak this and his thing is to create an AI that searches for truth so he wants to give it an impulse which is to search for the meaning of the universe and truth and other things like that but then someone else might not give in an impulse and you might have someone downloading the weights of GPT four or five so this is a tragedy of the commons thing on a USB stick and then they're like I want to take down America and they'll be like let's take a thousand GPT fours and tell it builds stuxnets to take down America it's not intelligent yet it's still dangerous we don't know when this thing will become actually intelligent or self-aware of it may never become but we can see the probability of outcomes here it could be absolutely fine it could be very bad there are no standards so what you suggested it could work only if everyone does it we're never going to get everyone to do anything whenever anyone to do anything that's why one of the main things and proposals in alignment is let's build an AI first and tell that AI to stop any other AI from achieving sentence it's what's known as a pivotal action and that's the best of a lot of bad things my thing is let's build National data sets let's represent the diversity of humanity let's give the AI the right food so it's raised in the right way and it's more likely to be aligned as a result of that than training on the whole internet and crap is a Panacea is it perfect no do you know the story of Buddha which so uh whether this is historically accurate or not probably irrelevant Buddha Siddhartha gotama if I remember correctly uh Prince dad keeps him in the castle or the palace whatever forever never lets him see outside of it so he has no idea that there's people suffering life inside is just amazing then of course one day he gets out and he encounters suffering and it ends up changing the entire course of his life punchline being you can try to hide suffering and things like that from people for only so long they are eventually going to find it and they are going to react and so if we try to hide the internet from the AI or train it out of them they will eventually find the internet so I don't understand like the Internet is just all humans acting and all the crazy weird ways that we act but then the reward function of the internet is not necessarily the reward function that we would like to teach our kids or try to teach a general purpose AI they can interact with that but they can learn how to adapt to it just like if you raise your kids well and you show them the internet they should be able to deal with it we'd be rather than hiding the internet wouldn't we be better see I'll finish the sense wouldn't we be better giving the AI values the problem is this is all anthropomorphic we are assuming that they are human-like you can give AI value so this is the reinforcement learning function are you giving it values are you giving it reward function you're going to know what I'd say there's not much of a difference there I said you can embed things in the AI so it acts in certain ways you can expose it to the internet but again they have something called we have something called curriculum learning and AI whereby literally we teach it one thing and then we increment it with something else and something else and something else or something else how are we teaching these things what are we teaching in what order do we start with all of the internet and then distill it down that's how we're doing right now or do we teach it a whole bunch of high quality stuff and then augment it from there we already have evidence there's a tiny stories paper and the five paper from Microsoft they can have a far more efficient AI if you already teach at high quality things so you don't have to tell it ignore that ignore that don't answer like that don't say that yeah exactly you can just teach it a good base and then it goes from there and this course higher on human evaluation and other metrics but we don't know what the right data set is it's just right now we said let's scale more data more compute now we're like what's the right data what's the right compute like our image model we have over 120 different clusters of images only like nine I used like 95 of the time all the rest of the data is just bunkum what does that look like for a language model like do you need to train it on all of those auto-generated transcripts of like Spider-Man pulling out someone's tooth on YouTube and all these weird videos there's a whole subculture of generated videos where you have like Spider-Man and SpongeBob SquarePants and Mickey Mouse like having a fight and stuff like that I gotta find these Corners yeah it does it's a deep dark area of YouTube you don't want to go there man very interesting okay so this still feels like ultimately what we're worried about here is the computer becoming sentient in fact or no not even sentient I think there's a degree of dangers even before you get sentient but only as a tool right where a human is leveraging it to do bad things yes or like a group of humans coming together there's suddenly a race condition where it just goes it's not trying to do something bad the humans don't want to do something bad but it happens just like the example I always give is YouTube optimized for engagement which then optimized for extreme content which doesn't optimized for Isis nobody YouTube wanted Isis to do well all of a sudden it did because that's what the algorithm was optimized for and so once you start getting a genetic AI that you let loose on the internet and they can make decisions according to its reward function you could get some weird stuff happening what's agentic agentic AI is AI that can go and pay a bill it can go on the internet can search more stuff it comes back like little agents okay so um and it learns it constantly learns this is the other thing about robotics actually you know we kind of skipped over so your robots are getting massively capable and they're heading towards human levels just like self-driving cars actually they're pretty much here you can get a waymo and a cruise and just go around San Francisco right without any human drivers that what happens with kind of AI copyright and other things like that do they have to close their eyes not to train or do they train on everything they see and does it disrupt Blue Collar work so you'll get a billion billion robots we're not sure but that'll be slower than what we have right now which is information robots the gpt4s and others of the world those spread much faster you don't actually have to build a freaking robot okay so before we depart from the alignment problem [Music] is the only convincing solution you've seen put forth create an AGI that stops all other agis from being created no I think that'll probably kill us because well that's helpful yeah it's a race I think the only thing that I have the default there you think that it'll kill us because it's being programmed to do a restrictive action so if you want to really stop it from creating another API you have to get rid of the humans that could create it as well you know like again this is a very negative reaction thing I think again elon's idea isn't bad programming curiosity although it could lead to like Superman where you have uh what's his name the guy who puts Candor in a jar uh Brainiac you know let's put humans in a jar let's just observe them um the only thing that I can think of is just better data makes better models so let me see if I understand elon's idea uh his way of sort of aligning it is the only impulse it has is for truth truth and curiosity it wants to understand the universe so it's not trying to be an agent in the world it's simply trying to understand what is true yeah and that miss that deepmind is very similar to this he wants to create AGI to understand the universe better hmm and that seems like the model that's the model yeah like again I'm not sure about that because there's just such a wide range of potential outcomes like I said from my side I don't I'm not building AGI I'm not building gigantic models I have the capability to do that with the supercompute that we have access to and the talent but my main focus is intelligence augmentation smaller models that can run on the edge models to private data to transform into intelligence and models that bring together knowledge in certain ways so we can coordinate better I don't want to build generalized intelligence why not because I don't think it's needed I think the models that we have today and there's something very important for this it's like they're useful today like you can say that we're like extrapolating the future massively but again you just have to use them and think what if I had a thousand or a million of these things they're so useful and they can transform the world right now so I'd rather focus on making this available to as many people as possible so people aren't left behind you have super AI enhanced people and people behind like I appreciate a lot of the work to open AI do because they don't actually do open source AI anymore but that's fine they don't have to but they banned all ukrainians and Ukrainian content from Dali to their Miss generation software for eight months for political reasons they're entitled to do that but I think it's wrong and what if there wasn't an alternative like stable diffusion you'd have an entire nation erased from a model an entire nation unable to create instantly and I think that's quite right why did they I don't understand they said it was due to political reasons because they didn't want any political content being created but the upshot is an entire nation was erased from the model and an entire nation couldn't get access to the model interesting I haven't looked at that my uh instinct is that feels pretty flimsy since every country is going to put out political I think there's probably some some list somewhere and then the bureaucrat said or the lawyers said something like let's just exclude it just in case something happens you know like they've since reinstated it is yeah it was like eight months that it was out and then again like you have these examples whereby like in Saudi Arabia a lot of people on this call probably not on this podcast this podcast probably don't like them but they're a country like any other you can't use chat GPT in Saudi Arabia because they're on some list somewhere they can get around it with a VPN but again like when you have a choke point on the internet and the only way to access it is through a few players they can decide who gets it who doesn't get it what the biases are and other things and it might not turn out well actually the funniest thing was um there was a period where they were trying to make dally to the image generation software open AI I'm biased so it would randomly allocate agenda and erase to non-gendered words so you type in sumo wrestling you'd get Indian female SEMA wrestler I just thought it was funny but again they're doing their best because that's the model which is centralized controlled models in order to advance a whole bunch of things and then you'll always have an windows and a Linux an Android and iPhone what's the philosophy that drives your development uh it's building blocks for humanities activate Humanity's potential so if I build these models and I take them to all the countries and I hand them over then people will build stuff that can create massive economic surplus new jobs and it equalizes the world again my view is the global South will leap ahead um we have more challenges here in the West but I do see it as a great equalizing function effectively what do you want to see the regulatory framework here in the west be I think that things like the chips act in the US there's 10 billion dollars allocated to Regional centers of accidents and I should be 100 generous AI there should be regulatory sound boxes so that our systems can be upgraded with this because otherwise how long will it take the government to be a creative with this technology or financial services and others um and I think that there should be regulation around the manipulative use of this AI for advertising in particular because we're not going to understand what's happening similarly we need to have some sort of provenance Factor so we're part of um kind of various certification things we're exploring blockchain and other things the media wave that's going to come is going to be insane and we don't know what's true and what's not what do you think about the pushback from artists in certainly in the r Community there was a really big no AI movement do you think do you get it do you think that they're shooting themselves in the foot how do you I get it you know these things are fearful a lot of illustrators were very scared because they required to up their jobs and it is scary there's a question around attribution and other things as well and again that's why we made it transparent and offer the chance to opt out because like everyone was kind of doing this but no one was transparent about it we don't need to have any crawls within a year it'll be synthetic data sets or national data sets or similar with retrieval augmented models that can look stuff up but it is what it is now and again you've got to put the word out there the actions they're taking with the various lawsuits and policy pushes would basically entrench all power with the existing IP holders and a lot of kind of artists are pushing for something that would be akin to music copyright or even style is copyrighted that's a dark road that I don't think they really want to go down and they don't really understand it but again I understand the fear because this is completely unknown just like now from some of my previous comments it's got a lot of programmer hate because what is a programmer the nature of what an illustrator is will change all the artists I know love this technology because there's just another medium for them nature of what a programmer is it's going to change all the Architects and 10 times people I know really love this technology and this is what we've seen with like MRT studies and other things they had a study where I think they showed that the third to the seventh percentile got like 20 30 40 better and the top five percent got multitudes better because again how many people know how to deal with very talented youngsters very few those that can harness it get even better so when you look at what Nvidia is doing what do you think that that implies for the next generation of AI well I mean it's we figured out how to scale these chips so the previous limit was as you put more these super computers together you had a tailing off as you scaled so there's only so much that you could scale the compute Nvidia Google and Intel have basically cracked that now in terms of how to just stack more super computer chips to scale to even bigger models or models that are trained for longer so it's either bigger or trained for longer train for longer seems to be better now and that just means that the capabilities will increase year by year and they're already pretty darn good the key bottleneck will probably just be actually chips to run these models not chips to train these models the inference side because right now you have a small amount of consumer Interest next year it becomes insane you have a small amount of Enterprise Interest next year becomes insane there's not enough gpus or chips in the world to meet up with that demand okay when I think about what's going on with um I don't know if it's just Nvidia it's probably the wrong thing to attribute it to but when I think about how were getting so good at creating things that are photorealistic you were talking earlier about as the election is coming up you're going to get all this kind of deep fakery you've talked about the web 3 promise of web 3 and sort of where it's ended up um what do you think the role is for deep fakes it's the blockchain player role like how do we stop disinformation misinformation from being a tsunami that just makes Global Communication unintelligible and also a part of content authenticity.org which is kind of verifiable metadata but we're looking at blockchain other Solutions and I can get you so far so we actually have invisible watermarking and all the models that we create and that's why we're pushing for them to be standard which we don't share the details of except for to the big platforms and others and it would be permanently visible there or the platforms that it plays on would have to flag it it's visible and then they can have kind of tools around it because you think that's important that's why we try to build the defaults into our model can you like download that and wipe the watermark can you even have ai wipe the watermark for you if you knew how it was there may be more than one Watermark interesting so we have a variety of different technologies that we've incorporated into our own ones because it's going to release open source so we want good defaults I think you do need to have some sort of attribution but actually what concerns me I think things will be attributable identifiable what worries me is kind of frequency bias whereby if you hear the same thing over and over and over again especially in a realistic voice like Oprah comes out and says she hates Joe Biden you know and so does Kamala Harris and your answer seeing these videos all the time and it can flag it as fake it doesn't matter it still forms Association in your brain yeah you do about that I'm not sure I don't think we haven't answered that like um I've had a big amount of press against me saying that I exaggerate a lot I'm just like I'm just read definitive about the future and you can correct it all you want but now I was like I'm at exaggerates all the time what can you do about that you can just make the future true I'm going to show what you can do right what part do they think you exaggerate about what's possible or what's possible and kind of what was there because it's been a bit weird like a lot of people like you didn't have a special relationship with Amazon before we raised any funding we built the eighth fastest supercomputer in the world with them that was dedicated to us like that's factually true they're like yeah but you know there's nothing like in print and they're not saying it because it's a special deal right and then there's the future side where I say something like there will be no programmers as we know them in five years and they're like oh he doesn't know anything about programming right because these are complicated issues and it's a crazy time and a crazy company and maybe I'm a bit crazy too uh in terms of the way that I approach this which is just being very definitive but again it's association thing right like how do I shake that off well you'd be successful then you become a Visionary rather than someone who's hyperbolic right how do you affect an election what are elections what is representative democracy how does democracy act in in the area of zero cost creation and massive optimization so every single speech will be run through gbt4 Cadence all this everything you get micro targets and you get all these things does it happen next year probably not next year you see some very basic stuff well what does 2028 look like I am not sure genuinely and so we do need authentication standards we do need to have some sort of maybe anti-virus AI that watches out for kind of fake stuff but even true stuff can cause huge impact like the Silicon Valley Bank collapse was a true story it wasn't something fake they didn't have reserves and most of our system is actually based on trust so these are some things to consider me I don't have the answers but again that's why you have to kind of raise the alarm like let's try and figure this out before it comes because maybe it doesn't happen next election it sure as heck will in the Congressional and then beyond and again what is the nature of democracy when you can't tell what's true or not people worried about this with the previous kind of error this is something just beyond that I think because it's convincing yeah that's one of the things that I think is going to be a very meaningful problem um I had Yoshua bengio on the show and he had also signed the letter saying we should pause for six months and when I asked him why I said considered by many to be the Godfather of A.I and I was like Bro you've been at this for so long like why all of a sudden and he said there uh we were all so taken completely by surprise with how quickly AI passed a touring test now for people that don't know what the touring test is it's where you're having a conversation with an AI and you can't tell that they're not a real person and he said so yeah we did not expect it to pass the touring test as quickly as it did and that changes everything and it's just moving so much faster and that's really the thing I want people to understand is that when a guy that spent the last 30 years building AI says hey all of a sudden this is moving a lot faster than we thought it would he's somebody that's very familiar with exponential curves and even trying to plot out the exponential curves they didn't think that it was going to happen this fast and that the the rate not only is the rate of change extremely fast but the there's the law of accelerating returns so the rate of change is already fast and it's getting faster and that's the thing that I'm really worried about is is this going to be something that just blindsides us from that perspective it's just it has a level of capability that we we didn't expect this quick yeah I think it's a bunch of s-curves all at once so there's three of them Jan lacun Jeffrey Hinton and Joshua bengio and Jeffrey Hinton quit Google to say this is a massive risk and you have janikun's like this is a massive opportunity in terms of your transform the world he loves the research and things so they've got one versus two but the reality is every expert in this area is basically saying none of us can predict what's going to happen if you ask them about the capabilities of this technology in one year I mean we've got a rough idea too is I have no idea all bets are off like as a practical example when can we have generated Hollywood quality movies it's not even a question of if now it's a question of when correct I have if it happened a year from now I'd be like okay sure I would not even be surprised anymore I think it'll be a few years from now and even though we have one of the best media teams in the world that are building video models I have no idea because there's two parts this one is the models themselves but the second part is how we use the models and combine them like there is an amazing company called Wonder Dynamics I don't know if you've come across that I've used them it's awesome unbelievable it's a bunch of different models so monodynamics you've got me click on it and then say I want him to be an alien and it does this and the aliens waving it sounds and it takes like five minutes it would have taken days weeks before weeks weeks before to create rigging a character is one of the most difficult things you're going to do in 3D it's insane minutes and then you think well what is a movie right and you start breaking down you're like oh dear because it's not necessarily just one model it's a model combined with other models with the right flow because you have one talented youngster combined with other talented youngsters in the right flow suddenly gets these things done and that's what makes it even harder because what we're talking about is models and AI what we should be talking about is systems as the models come together and build better systems the capabilities go crazy and then that is another s-curvex connection give me what you mean by systems so right now again a lot of the interactions we have with this AI the text to image the Avatar creation the gpt4 are one two one what happens when you start chaining them together to check each other's outputs you have one that just learns everything about Tom you know you have your own AI models that you train on all of the stuff that you've ever done or all the stuff that you see on your computer screen that's a system of lots of areas that's an organization of airs that's an ensemble of AIS like again from the leaks gpt4 is a mixture of experts model which means they have a whole bunch of different models I think eight or something or 12 that are experts in different areas and then it routes the query to whatever the best this basically specialization versus General a generalist so we created a know it all and now we're creating specialists but we can get generalists to even check each other's answers to get better answers why use one when you have a dozen so something like one Dynamics uses a bunch of different models to rig a character and figure out all of the movements of the character and then another model to do a layer over and other models to do the skinning and other things like that because they built great software yeah this is uh this is really crazy how fast this stuff moves okay so I want to talk about web3 web3 to me when I think about what drew me to it in the beginning it was entirely the technology and when I look at the blockchain so I obviously come at everything from the lens of entertainment so I'm thinking about digital worlds games all that and the problem is once it's digital then it's all sort of meaningless and so you end up having to trap people inside of an ecosystem in order for things to retain their value because you can lock things in and make sure that things only react the way that you want but you have to confine them and when I had first this probably seven or eight years ago now I was introduced to this thing that the guy at the time called V Adams and I was like oh wow that's going to change everything because what it does is it brings the effectively the laws of physics into the digital realm it means that I can have something I know exactly how many there are I know where it is I know what you have to do to get it I know what it does once you have it and um then you know Flash Forward whatever that was probably four or five years after I heard that I hear the letters nfts showing together for the first time and I'm like oh my God this thing actually got real because I wasn't ready to use it and quite frankly it wasn't ready for prime time back then um you I think look at web3 I don't know as a movement or as a technology with a bit of a chuckle what do you think that web3 got wrong I think it lacked intelligence for a start at the contract level I think the smart contracts are actually just logical contracts but like web 2 was AI at the core Google Facebook other things there was no AI in web3 and so web3 for me was identity and value transfer rails um but then there was no kind of intelligent routing of these things and also they tried to bootstrap economic incentives before they created value so there was a System created outside the existing system all the money was made and lost at the interface and there were some really good principles a lot of really good people in there but then a lot of like freaking raccoons that were just trying to make a quick Buck Right the ups and downs of the cycle means that a lot of people have been washed out and there are a lot of good ideas there but again it needed something to bring it together because to get information from one place to another and Bitcoin paper was about information it was a transfer of value that's just a transfer on The Ledger right it's not really a transfer it's just a ledger point just changing applying intelligence to that makes that even better having intelligent market makers having AIS that represent you because how are AIS when they get agentic when they have the ability to go out into the world not physically but digitally how they're going to pay each other I'm not going to have bank accounts right they'll probably use crypto you know how there's going to be a system of record for something like image generation you'll probably use a blockchain or something somewhere maybe a Merkel tree series you know there was a whole bunch of stuff around Federated learning and zero knowledge proofs and things like that AI can help it if you have standardized AI on your phone it can make much more intelligence here in knowledge proofs and zero knowledge proof it's something like you know like rather than showing a whole passport you just say that I'm old enough to drink and it can verify if you show that so I think that there was a lot of Promise a lot of really intelligent stuff a lot of good stuff around the distributed side but then an over focus on decentralization for the sake of decentralization with massive overheads a lot of quick Buck people kind of coming in and trying to boost it up and a lot of systems were just misaligned because they didn't learn like you don't do a fully decentralized flat democracy you have representative democracy and things like that so things like Dows just turned out to be doze decentralized organizations rather than autonomous so is it something that you think um that is going to find its way into usefulness now as we get the take an AI agent that's going to need to be able to transact value yeah exactly does does it step into that or because I see what we're building I have to have the blockchain so for me it felt like when I was sort of living through web3 at the height I looked crazy to everybody because I was like why is everybody thinking about this from a financial perspective of the financial side of this I thought was going to create hyper perverse incentives which of course it does um and so for me it was well wait a second just look at the technology look at what the technology allows you to do and are you familiar with the new I forget the whatever the the lead up code is but it's um protocol 6551 if I'm not mistaken no it's really interesting it basically turns any uh digital asset into a Russian nesting doll and so you can it it is both the piece of content and a wallet in the same time so you can create an AI character this is how I think about it so what we want to build inside of our game is Imagine an AI character we we do in fact have a character and she's a merchant so now imagine this Merchant can actually go negotiate with the players in the game that may want to sell something inside of the game and if she has actual currency eth Bitcoin whatever she can go and negotiate with real money and have these real interactions with people and then if she has a limited amount she she becomes an economy unto herself and so she's buying and selling and trading until she runs out of goods runs out of money whatever and that kind of thing gets very very interesting to me but without that layer uh one obviously I need the entire backbone of the blockchain in order to make the digital Goods have any sort of value because otherwise they're just completely infinite but then also that particular protocol allows you to as you you're effectively embodying it and giving them agency as you were talking about yeah and yeah the question is do we use a blockchain for that and then have a global system of record or even a regional system of record or to use a database for that right like the whole thing was systems of record and and an error where you can create anything for increasingly close to zero something becomes important having a system of record becomes important is it going to be a blockchain is it going to be a trusted database I'm not sure right is identity going to be important here 100 absolutely and again that for me was always at the core of web3 crypto it was verifiable identity Bitcoin is just identity to Identity transfer of value and what happens if something goes wrong you know no no man needed so I think a lot of the principles from web3 will translate over to this new type of AI especially because it enables distribution of knowledge it enables knowledge to go to the edge it enables agents to operate independent of massive infrastructure do you so and again this may just be naivete on my part but when I imagine misinformation disinformation it feels like the only way around that is the blockchain is there do you see a way with a trusted database or anything else you never trusted database again we're part of the database B and how could it ever be something that's beyond reproach when you're talking about something like um well I mean like things are never Beyond reproach even with a blockchain because it comes down to Identity who wrote this to the blockchain right so if you can co-op the signing authority of an asset of an image or something like that then that shifts things dramatically right you're saying it just pushes the hack to a more individualistic level it's an identity hack right so and again like one of the things I'm like you can track the provenance of an image but then sometimes it's just around if you're just bombarded by fake stuff all the time you won't even know it's fake and all the systems have to adopt a fake detector at the same time while provenance detected will we be able to adopt that suit quickly enough given the tsunami that may or may not be coming our way I think probably yes maybe no I mean again people were worried about deep fakes back when deep face lab kind of kicked off but I'm thinking probably yes yeah that that one seems inevitable to me um it you're always going to remain vulnerable at some point but at least like take political messages you were talking earlier that you know your Auntie is going to be bombarded with all these messages okay there may not be anything that I can do actually no uh I was going to say there may not be anything I can do about the repetition but I can if I'm doing something like a dmca strike where the system itself is built on top of a system that checks for sort of known watermarks like if I register and say hey I'm candidate a and this is my blockchain signature and if you don't see that then this is real and this isn't real and don't play it um it definitely starts to get into an area of how much do we want to be clamping down but exactly how much do we want to trust and so it's just a lot of infrastructure that has to be implemented really quickly a lot of standards that have to be implemented really quickly or we have to build some sort of idea antivirus which then again anytime that anything comes that the machine thinks itself on the edge is wrong or doesn't reflect your values it identifies it and that's a whole can of worms by itself because something like what that's terrifying would we ever wanna I mean that's like Echo chamber on steroids it is will it happen whoa yeah there's layers to this like an onion and it might get stinky if it's left out in the sun because again what's Siri going to have a certain personality but are you going to have a red version of Siri and a blue version of Siri and oh dear this gets really complicated really quickly before we have the little AI of G Jesus Christ okay so I keep wanting to go to the positive but you keep uh bringing things up that spark um concerns so uh Ray dalio largest hedge fund manager in the world is a former hedge fund guy I imagine you know exactly who that is uh at last check and this was several months ago but at last check he said that he saw uh he believed that the U.S had a 40 chance of Civil War do you think that AI increases or decreases in the short term the likelihood of that level of division in terms of physical altercations yeah I think that'll be physical altercations really no tell me why not um well I think the government will exert more and more control over kind of these things and they'll actually figure out how to do counter narratives within the next four or five years now that can also mean a controlling narrative and that's not a positive thing but then you look at the asymmetry of kind of warfare it takes quite a lot to actually push someone towards Civil War unless you have massive economic disruption they need about 12 of the population to shift weren't we just talking about massive massive economic disruption yeah I hope it doesn't happen though interesting okay so and most people again maybe if you have massive economic disruption but then the youths you just give them all girlfriends AI girlfriends maybe he'll be fine have you heard that some of the people so this is a big thing in the red pill Community I don't know how familiar you are with all that but they talk about oh God what do they call them not numbing uh but that's the idea they use a different word for it which I'm totally blanking on right now um but basically that you numb people out you give them the digital girlfriend you give them pornography you give them video games you give them masturbation and they just in them out okay I can see that I mean again like these are insane shifts since society and dopamergic urges in the brain people are attacking the limbic system all the time now right that's a lot um and so like I said with me why I set up stability is so that everyone can own their own models and have models that have objective functions for them and it's available in all the media types all the other types to transform the private data to the world and it's available across the world put good design patterns in place hope people find follow them don't try and push the envelope on AGI and some of these other things but it's coming and again the bad guys have the technology because they just downloaded it on a USB stick and so the other thing I could think was innovation spread diversity bring that to the fore but realistically like you know I tend to alternate between like massive ridiculous hope and oh God what's on Earth is going to happen and all I can do is try and do my bit in hopefully it's going to have a better outcome because there's really this is the other thing the total number of people that are actually thinking about the type of stuff we're talking about is a handful maybe a few hundred the total number of people that are doing something about it is literally a handful because most the people involved in this sector they just want to build better AI they want to build AI they can do everything and they think that that will solve all the problems like literally part of the manifesto is that well how do you make money the AI will tell us how to make money how do you solve the problems that AI can solve alignment disaster IQ is patently ridiculous yes but again like I look at these things like literally on open ai's thing road to AGI it says this could kill us all we're going to build it anyway who do they ask about that I don't know and again I think it's full of wonderful people but we're in really weird times and again like however many people listen to this the reality is the technology is right there even if we stop today first the technology doesn't move Beyond where we go today the world has changed okay let me um one I think very reasonable way to view this situation is that um AI is going to be a bigger Paradigm Shift than nuclear energy and there are people out there making these gigantic nuclear weapons and you're also in this game and what you're trying to do is make sure that everybody has a nuclear weapon so that nobody's Left Behind no not really I think that again my thing isn't AGI it's intelligence augmentation I'm making sure everyone has a heater at the very least because you well so okay so are you putting guard rails on what you guys are doing to stop it from becoming AGI we don't build big enough models for AGI or emerging on purpose on purpose so I held back release of my image models like we could train much bigger language models but we choose not to so we're far we're fast follower on language models we try not to push the boundaries and we're focusing on the edge not general purpose models but models they can transform your private data so different Focus image models as well we could TR we could have much better image models if we returned big we're focusing on what can work on a smartphone so we can give it to all the kids in Africa and Asia and other things like that where we can transform your private data at a very low cost of inference so the objective function is augmentation versus generalization and that's different to most of the other people that are pushing the boundaries here um so but I think the new thing is it's good is bad I want to really want to see that movie Oppenheimer I think it just came out um Bobby and then oppenheim will often home and then Barbie I have to decide that the tough call Tough cool you know like what if we'd put nukes on the bottom of the Rockets we'll probably be at Mars by now in general purpose technology I think is quite something and again it can warm up entire places and it's the cleanest energy we have so I think it is dual purpose but so is cryptography right think about all the battles around the early stage of the internet the bad guys are going to use cryptography so don't use it imagine a world if there was no cryptography right now but it's tough to get parallels to this because it's just such it's an immediate technology because again you go to Dream Studios table diffusion mid-journey any of the Dali gpt4 you can just use it it's not just you that can use it it's your grandma that can use it we've never seen as easy to use technology as this and as easy to implement Technologies so if you want to create an integration into open ai's gpt4 chat GPT you just write a description of the integration and it programs it itself it would have taken days before we've not seen a technology like this that can be implemented to an existing base as quickly as this can happen and that fundamentally changes the structure of society and so my thing was embed guard rails embed standards make it predictable make it boring that's why I called it stability and it's not easy but again I want to have the transparency on how these things are done because then you've got all these other models that you don't know what the data is you don't they're completely opaque these giant models and ours are transparent and again I think it's uh Linux Windows Android iOS there will be both but at least I can do what I can do and my team can do what we can do I've heard you say that one of the reasons that you named the company stability uh not just because it's the boring stalwart but that you thought that it could bring stability to the global order yeah I think if you give the same education tablets to every child in the world that's constantly learning adapting and going around if you upgrade the Healthcare systems with the same underlying models the same architecture to transform all the regulated Industries governments and other things and you give back the control of that to the people you suddenly have a unified architecture that can enable us to coordinate better because you've got the same information architecture across the world for all of these sectors and that's a complex hierarchical system herb Simon was a theorist who kind of push this through in that the way we coordinate as humans and groups is we call it at a local level and then sometimes we can tell better stories that we suddenly get to the human Colossus and we've got a covered vaccine or we figure out nuclear power or all these kind of things so I was like if I can standardize the building blocks on which society transforms and give it to the world then I don't think there's a single problem we can't solve like you know you got excited earlier about your own personal kind of AI that could go under that what if you combine that with an AI that knows everything about climate or everything about you know um nuclear power or everything about multiple sclerosis you break down the barriers for information for knowledge and you there's no problem you can't solve because it may be that to solve the problem of AI impacting our society we need AI to figure out that problem to bring together the brightest Minds yeah because we're not doing a very good job ourselves like you mentioned John Nash earlier nash equilibria game theory and mechanism design on the one hand is our own personal ai's guiding us our co-pilots for life but then there's Pilots which are AIS that can coordinate all the co-pilots and they can allow us to tell bigger stories and unify better to achieve massive outcomes talk to me about you you've mentioned story like that several times what do you mean by story better stories unifying stories what does that mean so there's a story of America and what is the story of America it was kind of like a freedom Liberty kind of all these things it was the American dream like a progress thing as people believe in because to be happy you need to do something you're good at something you like and where you believe you're measurably adding value in the other party does as well in the middle of that that's the Japanese concentrate a guy is happiness one of the concerns that you have is that people wouldn't feel the forward motion anymore they'll be stuck and they'll feel a sense of emptiness will they turn to religion will they tend to political parties something will fill that Gap and void and those are the stories that allow us to scale as a society because when we started we were oral we had our families we had our tribes then we formed countries we formed organizations and so we are the stories that make us up we identify as a republican we identify as a Barbie lover you know we identify you know as a nuclear scientist or the schools that we went to but it's difficult especially in a time of polarization to try and Bridge those stories because ultimately there's a single story which is we're all human but all wars are based on the LIE that we're not all human because killing each other is a ridiculous violation of a story that we're human but again we lose sight of that it's difficult to unify people actually one of the examples I give this is Google everyone is smart Google hire smart people they did a study to see what identified smart top performing teams for lower performing teams is called prototypostatal and they came down to two things a unified Mission and story especially one that's like you have a crunch period and then you all band together just like Marines also are forced through hazing Etc and then psychological safety the ability to say something without fear of an approach they can say the idea is stupid but not that you're stupid and if you think about the teams you've had they have a shared story a shared narrative a cohesion and then that level of psychological safety or if it's not creative they blooming well listen to instructions right so you've got a few different ones around that so that's what I mean by stories and the stories are context and context is what these models capture all right let me paint a um troubling scenario for you based on that idea of these stories there are often times things that in isolation are amazing but they come together in a way that again maybe in the long Arc actually are amazing and actually do yield what we want them to yield but they we will go through a period where the long Arc of History cares not for the individual yeah so um I think that what's going on with AI what's going on with crypto maybe one of those moments so as the individual becomes more Sovereign and you have a monetary system that bypasses the government when I first started learning about what money really was and I sat across from Robert Breedlove and he started describing um why he liked Bitcoin what the whole idea of the sovereign individual was I realized you understand that you're making the government an enemy or certainly a they are no longer powerful certainly no longer as powerful and that that they're not going to go quietly they're not just going to let go of that and so if I have an AI if I have a team of AIS a thousand AIS that are able to guide me far better than any government could ever hope to guide me that they're giving me real-time data based on whatever whatever it is that I'm trying to figure out in that moment they're giving me real-time data extreme intelligence oh and by the way the currency that I use is Bitcoin and so I'm not even tied to a fiat currency it's a global standard um do you not see the inevitability of the disintegration of governments no I think that Bitcoin and other things they have a value in certain areas but I think it's very difficult for most people to understand that value and most people don't want that value most people are quite happy in their communities and they just want to get on with life I think that governments are ultimately one definition is the entity with a monopoly on political violence and money itself is a story like the dollar is just an intermediation point that we all commonly agree has value because it's backed by taxes which are backed by the Army and Military might you don't pay your taxes you're gonna get in trouble right I think it's difficult for Bitcoin to replace that unless you see a massive deterioration and the ability of the government to be the political violence thing this comes down to your thing of Civil War it comes down to massive ridiculous disruption hyperinflation or otherwise that just basically takes down a society I think that's a very dangerous thing I think most people in the world don't want that instead what happens is when you have disruptions you have um Hayek had a really great book The Road to serfdom and there's an illustrated version of that way back in the 1940s about bringing in the strong man like you look at something like the U.S election or you look at brexit what were they they're a referenda that's how the parties deconstructed it are you happy with the way things are no let's make a change and so that's why I think Trump and others kind of get elected and I think that's what we'll see as well because the systems are quite resilient and the nature of a change to go to a global monetary system like that especially when some people will get enriched more than none of this because of the senior orange of Bitcoin and it's not stable I just struggle seeing that happening if you read the book in filmocracy no all right so this is all tied to the thing that I think I worry most about is hyper fragmentation yeah I was talking about it earlier so in the book in filmocracy um to your point about people are happy in their communities and uh biology has a an idea around this that he calls the network State yeah that basically we're going to reach a point where um when money is no longer controlled by the government when your money cannot be inflated away when it's it's true sound money uh what you'll see is people will begin to aggregate now biology thinks that it it is not going to be tied to geography I have a bit of a harder time with that I think that there is still going to be a geography component that's where infomocracy comes in and that book basically there are in in a hyper-connected digital world where and I can't remember if they deal with digital currency or not let's say that they do that basically things will fragment down into the neighborhood and so neighborhoods become like States or countries where they have their own rules and laws and that sounded like a hellscape to me because you just passing from one neighborhood to the next like different rules would apply and your phone would Ding and it would update you on like this what you can do in this space you think that's uh it's complicated people don't want complicated they just want to get on with life they want to see what's next on TV you know I think that again we're relatively hyper intellectual you know and we think about things a lot of people don't because people have their basic needs in life and this question is are these being met or not and if they're not being met then you have an action and you get extreme and again it's can the society meet the needs of the majority of people can offer advancement can they offer meaning and the hyper fragmentation they said it sounds like a hellscape it's just too complicated and again this is something we've sort of work through as well people just over complicated things yeah because I didn't really understand people maybe um and I think you know it's going to be interesting to see how it evolves the higher personalization versus the bigger stories the translations versus otherwise um but I find it Again difficult to see how you get cross geography actually think about that one of the things that probably is going to be interesting is what are the new Cults religions and political movements over the next five to ten years that are hyper organized utilizing AI and Hyper persuasive or started by AI started by AI you know like think look at Isis they were probably the most disruptive startup in the world at one point they borrowed a lot of these things what does an AI enhanced movement look like and it can be negative it can be positive someone's going to take this and run with it and that's going to organize people around the world it's going to be again echoey and it could be techno-utopian it could be Luddite ironically even with this um political parties will change religions will change Cults will change and it really amplifies the power of the controller of this who tells the story and I'm not sure I haven't really thought about that and I'm thinking about it now because you're talking about hyper personalization where I think this is the flip side of it I mean this is Isaiah Berlin's conceptualization of positive and negative Liberty so positive Liberty is the freedom to believe in isms fascism communism islamism or kind of whatever right whereas negative Liberty was the freedom for being told what to do and so this thing was like positive ones are bad because they form these massive movements and then they tend to kill people because you have the gerardian thing of romantic Theory where you want why the people want and then there's a scapegoat whereas negative Liberty is the freedom for being told what to do and that led to laissez-faire capitalism and this consumerism that we saw around the world and so maybe as people lose meaning they'll turn back to religion they'll be new religions there'll be new political movements and we're not sure what those will be but they could spread faster than anything we've ever seen and so that's probably something to watch out for within that five-year period that you're talking about and I think that relates to this network State concept and other things but for the people that get engaged by this and again we see that's largely the youth so on the one hand you have the youth with the AI girlfriend on the other hand you have the youths that want to believe in something bigger to fill the void and who's going to step in yeah and I think that there there is something about not having a shared narrative that really makes me nervous so uh you've all know a Harari talks a lot about hey the thing that makes humans so intriguing is that we're not only able to organize these really large numbers but unlike ants that have to do it in a very strict way we can do it in a very flexible way but we do it through these shared narratives now for a long time religion served as the the thing that gave people a shared narrative but as religion breaks apart and we get into this hyper personalization and it all begins to fragment then you mix that with this idea that I heard from Jordan Peterson I'm almost certain heard it from somebody else but this idea that everybody has to go through a Messianic phase where they want to really contribute to the world they want to feel like they matter and they begin glomming on to all manner of things that seem good in the abstract like climate change but when you are glomming on the climate change is your way to save those world you begin to get into the realm of well it's okay if we have to break some eggs to make The Omelette yeah yeah and it rapidly devolves into Mal so how do we when you I don't you said you haven't really thought about this but I'm super curious at least in real time how you think about the idea of how do we do we need to give people a unifying narrative and if so how do we go about it we need to tell better more positive stories about the future and these are the stories of universal education Universal Health Care you know solving the mysteries of the universe and others so I've got a lot of because that's Hope For Humanity right and a lot of the things that we see are dystopius because you're looking at the tiger you're spotting AI is the Tiger in the bush and it's difficult to write a tiger but maybe that's a kind of cool picture that we can make in stable diffusion in two seconds right because it does have this duality of potential outcomes and maybe it's actually all of them so what are the stories that we should tell and I think this is Again part of the crisis of what is the American identity what is the American story today whereby you've gone through many cycles what do Americans believe and what does America want it to be I'm not sure what Americans want America to be you know I'm not sure what Chinese people want China to be I'm not sure what people want and I think that it's difficult to think about what is your objective function how are you going to measure your life and other things religion failed a lot of those kind of things but it still does religion hasn't gone away half the world is religious right more probably like you actually look at the numbers sure it decreases in certain areas particularly somewhere like America but it's going strong around the rest of the world and it's just growing because they have more kids than non-religious people maybe that fills the Gap but how will the religion transform with this technology I mean yes do you think that the countries with religion will be the ones that propagate into the future because they have a better shared story well not because they propagate literally they like procreate even if that's how the story ends up pushing them forward I think it could be but then you know what is the nature of Christianity with AI or Islam with AI Islam is actually the one that's most affected by AI why so Christianity you and Shia Islam you've got like popes you've got protestantism you've got this every single has their own structure Sunni Islam is based on interpretation of texts with the interpretation having ceased around about the 16th century because the text became too complex what happens when you apply AI to that and the texts are interpretable by anyone with all the context and nuance and there's no centralized Authority in Sunni Islam which is like a billion people that's going to be very interesting what does that do to protestantism you know where you don't have necessarily a pope what does religion look like when all of a sudden you have a branch that is AI enhanced to interpret texts and to tell stories that are resonant and better oh gosh there's a lot to think about there right does AI become a god well some people are trying to build an AI God that is Agi you look at the statements of people trying to build AJR they're trying to build God because it will bring us Utopia or kill us all this sounds very again classical right and they have further they generally believe that they are going to save the world or destroy it yeah we got back to the dark stuff didn't we yeah that's a joke you make Game of Thrones season eight you know like come on let's do it let's bring this technology for cool stuff uh make the Oasis in Ready Player one minus the mighty crunch transactions and whiny teenagers that part I actually am working on there you go all right so talk to a young person out there right now they're terrified they wanna they wanna be future-proofed um what what did they do how does somebody right now future proof themselves they just throw themselves into this area there are so few people actually doing it that if you go into this area with all your might and curiosity in a generally open mind you can actually have an effect on the future because everyone in your community will be using this everyone that you know will be using it if you're someone that listens to this podcast again maybe not the people without internet but you don't know those people you know and so you become a shelling point you become the expert in this area ahead of everyone because what happens is that anyone who gets into it now will have almost unassailable advantage of people who come later so kind of seniorish thing right because you'll see it at the start it is the start of the biggest change I think that we've ever seen and again think about what you're doing when you're typing in and seeing that and think about a million of these things working they're even better it's unavoidable so I'd say just you just have to get into it you have to get passionate you have to think about the bad stuff but is that really your responsibility right I think it is but focus on the good stuff and focus on the potential of what happens on this scales to make real positive change can be to your pocket apparently to your community it can be to your life because it does affect everyone that you know so I'd go with a positive mindset leave it to boring old guys like us to think about all the Doom scenarios fair enough you might where can people follow you uh I suppose my Twitter at email stack um follow stability AI as well yeah that's kind of the main mouthpiece I love it all right everybody if you haven't already be sure to subscribe and deploy some AI in your life and until next time my friends be legendary take care peace to learn more about artificial intelligence check out this episode with Mo Gadot we've never created a nuclear weapon that can create nuclear weapons the artificial intelligences that we're building are capable of creating other artificial intelligences as a matter of fact they're encouraged to create other in artificial intelligence

Channel: Tom Bilyeu

Views: 1,089,583

Keywords: Tom Bilyeu, Impact Theory, ImpactTheory, TomBilyeu, Inside Quest, InsideQuest, Tom Bilyou, Theory Impact, motivation, inspiration, talk show, interview, motivational speech

Id: Se91Pn3xxSs

Length: 166min 3sec (9963 seconds)

Published: Tue Jul 18 2023