Eric Schmidt on Navigating an AI-Enabled Future | TransformX 2022

[Music] we are joined next by Eric Schmidt co-founder of Schmidt Futures and former CEO and chairman of Google and Alexander Wang founder and CEO of scale AI for a fireside chat moderated by Derek Devine Tech reporter for the Washington Post please welcome to the stage Eric Schmidt Alexander Wang and Garrett Devine [Music] tonight [Music] hello how's everyone doing it's funny I didn't actually see this room so I was like how big of a stage are we coming out on but this is great um very welcoming um yeah this is cool to do this kind of thing in person I'm sure this every single person comes on stage is like wow we're in person it's probably even that is getting a little bit um boring but yeah it's thanks everyone for coming out and um braving the real world to kind of have this conversation and make these connections um you know I think we obviously have Alex and Eric here and they just you know have really interesting experience and perspectives um and but I kind of wanted to start a little bit so I I've been covering tech for like six or seven or eight years or something and when I started it was like aiai and every company was an AI company and we kind of just I kind of just stopped writing about it because I was like 90 of you are talking nonsense a lot of this is not real and you know I was you know you can follow the the kind of cutting edge of the science and a lot of that was happening at big companies like Google um but there was some interesting things happening there but in terms of the applications in terms of people making money in terms of business models being completely transformed it you know most of what people were pitching me it just didn't feel real and I think you know the last six months or even six weeks like some of that feels like it's really changed and I think Eric you mentioned Cambrian explosion as like you know a way to kind of look at maybe this moment in AI so I mean I think the first thing I wanted to ask you both is kind of can you give us sort of a tight description of this moment and you know from a business perspective where are we right now kind of you know what is the opportunity what is different about right now than maybe a year ago or five years ago or no Alex you want to start yeah I think um I I 100 agree you know I I uh I do believe we're sort of in the midst of an AI Renaissance and it's it's true if you look at any metrics if you look at sort of the broad usability of the technology and how many people are actually utilizing basically The Cutting Edge AI in in record time you know I don't think there's ever been sort of a precedent of sort of the time from AI breakthrough to it being in the hands of millions of people in the same way that stable diffusion and Dolly too and these these sorts of AI uh image generation models have enabled and I think there's there's a new we're entering a new world where the malls themselves are platforms you know this is something that actually Eric I think you and I uh talked about years ago was that like Hey we're going to get to this world where the the models themselves are platforms but what that means is that now you have you have people building on top of them in in fully new ways and and sort of um you know there's sort of this concept of for a while AI seemed like a application technology and I think we're entering a phase where it's a sort of underlying platform technology and I think for many people in the industry that seemed intuitively like it was it was going to be the case and and that we're going to get there but we weren't ever in a state where you know we had generally usable algorithms and generally useful algorithms and so it's incredibly exciting and then I think from a business perspective I think we're about to see sort of this next era of of sort of explosion of productivity from economic perspective explosion use cases so I think it's going to be a Brave New World over the next few years well let's start by congratulating you for scaling a company right thank you sorry for the pun we uh not the first time I've heard it I'm sure um the I guess I've I've done this so long to me this is just another example of how Tech works and every cycle it feels bigger faster broader it's exactly what Alex said and the what happened in the last five years was you had basically Transformers took over or you know architecturally and the combination of that and these extraordinarily powerful gpu-based architectures tpu-based architectures and these very large models have sort of now given us that next opportunity to platform on the platform I think the diffusion models are particularly compelling because a simple rule is that AI used to be able to analyze patterns but if you can analyze a pattern you can generate something so you can analyze the pattern and you can learn from it you can generate it well that is a very profound shift in Paradigm for many many people I mean if you're sort of running a business that is not obviously in AI or you you know you didn't start your business thinking about AI I mean do you actually should you be thinking about this and how should you be thinking about that in this moment this is what you do yeah yeah I I well I think now it's it's a really really critical moment for companies businesses as well as Nations to sort of uh really sort of realize this speed at which this technology is moving and and fundamentally embrace the fact that they have to they have to reinvent themselves in the new paradigm or they're going to be left behind and someone else is going to do that reinvention and it's worth noting that every incumbent says this and very few incumbents do it successfully so I was trying to figure out where the really exciting stuff is aside from the big tech companies sorry for my BIOS and the answer is the startups right the other large companies that are not the big tech companies are not doing this stuff because they have not been able to incorporate these new modalities platforms data models and so forth into their existing data flows for whatever reason including for example that they don't know where their data is or it's not normalized or it's not labeled in the right way or they can't get to it or the data center burnt down and they're not on the cloud or something stupid like that um all of which is possible and it's it's not for lack of understanding I think you can talk to Executives at these Enterprise and they know that data is their fundamental blocker to getting to an AI future and so they know all the problems the issue is you know where does the urgency come from and I would say I would argue that now you should be the most urgent uh that you've ever been because you know you just look at the pace of progress in the open source community in the AI Community among startups and it's it's Breakneck there's literally okay and if there's this weird thing that's happened which is that a lot of stuff was invented in the key universities yeah and companies like Google essentially acquired those teams right as either through startups or by hiring the professors or The Graduate students essentially all of the combinatoric Innovation is now occurring in these large companies or very very well-funded startups right that's not the norm and so what you're seeing now is you're seeing this enormous commercial interest in the big companies and we can go through the specifics but you also have the open source Community following very fast so one way to understand is the open source Community normally would be ahead but in this case it's catching up but the Gap isn't that big yeah I mean I want to talk a bit about that because I'm you know we were talking a bit about the civilian AI event earlier this week maybe people in this room were there as well and there was you know some direct comments about you know the big tech companies and there's also you know in my world which is like you know essentially we're maybe the haters were the ones who are saying well isn't this bad or isn't this going to cause a problem and we ask those critical questions you know a big one that we're talking about is you know proliferation access and you know different Google when they did their thing and when the big tech companies do their thing they're careful about what they release what they don't release largely that's for business reasons but there's also you know is this a technology that we want anybody in the world to have immediate access to and then you have open ai's model which is sort of to do that a little bit more but also to be careful about what explicitly they release to the world and now you have these more much more open source movements that are saying look like we were going to trust the world and we also don't believe that commercially these things should be controlled by the few they should be controlled by the many so I mean this is also a big conversation that's happening right now and I mean I don't know where you place your company in that or how you think about that yeah well first off I I think it's super interesting I think a genie is out of the bottle right I think that uh in in some part because of these open source efforts we're in we're in a world where there's you know we're not questioning when AI is going to get to you know Mass adoption is or or if it's going to get to massive option we're questioning when and uh and so I think that's the first major paradigm shift I think our view is that uh even with the open source technology there's still massive gaps and there's still uh parts of you know the world like our the United States national security Enterprise um large these large Enterprises that uh really have not yet adopted much of this technology you know there's these massive swaths of the old economy which require access signal this technology uh as urgently as possible and I think it's it's necessary for you know our view as scales how can we help bridge that Gap and and bring these like breakthroughs to those locations I mean it's it's sort of a related conversation but Eric I know in the book that you did recently there is concern about proliferation about Bad actors having access to this technology so I mean how do you kind of square that risk with the need to get this out to more people and democratize it I think Alex got it right I had hoped three or four years ago that the really powerful stuff would stay inside of industrial labs and that the people leading it would restrict it for example open AI did that with gpt2 for a bit but I think the sort of cultural pressure and the the way sort of the the I don't know describe it the liberal Democratic thinking uh causes this kind of sharing and it's often done with a very naive understanding of how these things will be used so you can be sure that the people who invented Vision systems did not think that their Vision systems would be used by the Chinese to track uyghurs right right it was not on the feature list you know five years ago they were building this for this and so if I were one of those developers I'd be embarrassed that my technology had been misused I guess you could say the same thing about the telephone but the important thing is it's naive and I want to be very clear it's naive to assume that this technology is not going to be used by evil people right to somehow trust the community well unfortunately the community if it's a large enough Community there is at least one bad actor and let me give you a couple of examples a bad actor could easily weaponize the generation of malevolent viruses using basically the the Transformer tools that that are now so popular plus a large database of Biology another bad actor could easily use these tools to do highly targeted offensive attacks on specific individuals a bad actor could use these to generate very manipulative disinformation the change of society to change an outcome to cause a riot to cause people to be killed right so I'm not suggesting we shouldn't build it but I would like people to say we're building this amazing platform comma we need to figure out a way to keep this under control to detect it to police it or to monitor it right and relatedly you know um a good friend of ours Jared Cohen I was talking to him about this recently that I think technologists need to go undergo geopolitical training just in the same way that they go through other trainings for uh for their job because you know uh he he mentioned this story around Ian Goodfellow who's the inventor of the uh again the general adversarial Network which were the precursor to a lot of the the diffusion models and image generation models we have today and he uh you know obviously this technology is very exciting it's very interesting it's very cool he brought Ian to speak to former Russian and Chinese intelligence uh analysts and he and he had he asked them you know what could uh this these image generation models be used for how are the Russians going to use them how are the Chinese going to use them and what are the the negative externalities of those and I think that level of understanding and frankly awareness is really necessary as the technology continues improving and getting better I mean so you have government clients right at this point I mean so like do you when you have conversations with them I mean and whether it's the US government or maybe other ones I mean are you hearing these concerns are they saying yes sell to us but don't sell to XYZ client or state or like how do you I mean you're having these conversations right now that are business that are strategic who do we want to work with who do we not want to work with like how like I don't know if you're able to provide any examples but how are you dealing with those kind of tough conversations right now or are you even having them you know I think the first one frankly was was our strong commitment to United States national security um and uh and we made that decision because you know I grew up in Los Alamos New Mexico where the atomic bomb was first built you know the Manhattan Project which I think is a very important lesson for us all to sort of rock and understand um and uh and it was this sort of clear technology breakthrough that had very very meaningful implications for uh sort of global Affairs and global events over the following decades and I think that AI technology is a very similar similar kind of technology has similar implications for world events for geopolitics for frankly the the sort of Battle of the Free World versus authoritarian regimes and uh we made a strong commitment this was years ago probably back in 2020 when we made the commitment to work with the United States government and and use our technology for a strong defense you know Eric also has been very deep on this issue led the the defense Innovation board among other organizations and sort of has been a leading voice in this topic of utilizing AI for for National Security well I think that the situation in Ukraine is a really good reminder that we collectively need to be in favor of democracy um and when you look at what's going on in China and we get into all the details of what's going on China and Russia um we really want to be in democracies so let's start with that so what we're doing had better be consistent with the principles of democratic Rule and Leadership and voting and all of those kinds of things so I always worry when they're not in alignment to answer Alex's question about his company for the other companies you would say pretty much everybody now has an AI ethics group and what happens is that there's some some question comes up and it goes to the settings groups and they decide most of the companies that I'm aware of have they try to not be on the offensive side and they try to be on the defensive side of National Security that's that's how their AI ethics group come out the only problem with that is that in in Conflict uh defense and offense can often be the same so this is these are examples like missile defense systems get very tricky because of that issue um so as part of my work with the military the we we rode in the military adopted the U.S military on AI ethics for the U.S military and so which I'd encourage everyone to read because it's all it's all in public so I think it's worth saying AI is important enough that each of us need to have our own sort of AI ethics rule our corporations our universities and our governments need to do it and I hope that they're consistent with demographic Democratic principles right I mean you know like are we in sort of a space race specifically with China when it comes to this technology or is that idea a little bit too Zero Sum and maybe even self-fulfilling like are we are we in this point where you know there's no going back you know it seems that the hedges are going to go a little bit higher and higher with technology or you know should we kind of pull back the clock 10 years and say hey well technology can be a way to extend American culture and American values and you know maybe we should be careful about how closed off or maybe even how militant we're getting about some of these conversations I think it's unavoidable I I mean I I do think the hedges keep keep going up part of that is inherent to the technology so our one of the big breakthroughs of artificial intelligence of recent years is sort of this realization that scale really matters scale of the company but scale I mean it's a good name right so it's a good name it's very relevant um but the the sort of amount of data you have for these systems the amount of computational power the size of the models you know all these things really really matter for the power of your AI systems and that setup you know it's almost like how the size of your uh nuclear warhead stockpile uh is is a is a dimension upon which proliferation will uh will persist you know this is just another example where you know the models are just going to keep getting bigger and bigger and bigger and do more and more incredible things and more and more uh sort of innovative tasks that have the potential for both good and and great potential for harm and so this sort of race with China or Russia on on AI I think is someone want unavoidable partially also due to the values misalignment that that Eric was mentioning you know we in America were going to be thoughtful about using AI for Democratic Values but those same values are not present in how China is going to use the technology or Russia is going to use the technology and I you know we need to just be real about what that might mean for how the technology will proliferate I think the correct word is exactly we've obviously worked together on this for a long time and using the same words literally he's exactly right when he says race it's a race not a war right Wars are different we understand what war is yeah so because sometimes people mix those up right yeah well again yeah study Ukraine to understand I was just there yeah the auras of a real war and and I say this with great respect to the military leaders that are in the room um the so so it's a race for a prize that's important and that race is to set standards for Global Information flows and Communications around the world in that race we the United States have managed to lose the 5G race definitively which is a source of great frustration for me every hour and minute I'm on the freeway an example is I'm in Ukraine and I'm on the train you can't fly there and we have a star link on the roof of the train and I have 200 megabits now remember that the next time you're on an Amtrak train um but getting back to the things we've lost we have a low Bar for our public infrastructure in this country yeah but why yeah this is the wealthiest and most successful country on Earth last time I heard about it which was in grade school uh which is a long time ago for me so look the important thing is um in this race China is a very very well organized competitor they have a model which is called civil military Fusion what they do is they pick national champions they give them an enormous amount of money um this is not this conference is not about Quantum but Quantum is now getting three times five times ten times more funding in China than it is in the west right and we know it's just at the beginning of of its implications all sorts of National Security issues so the thought experiment to think about is let's imagine that in five or ten years whatever time frame you choose we have a system which is partial AGI you know it's capable of setting some objectives it's interesting and so forth how would you feel if it was a Manhattan project with your background but one run in China right you'd be pretty upset right we want to be we want to make sure that if there's any chance for that kind of Technology we want it to be built in the West in control by the West I'm going to go to questions in a minute so if people want to think about that but the one thing I wanted to ask before that is you know then you just sort of describe China's industrial policy about how they're approaching it and you know the state is very involved in that to the point where they may be telling Business Leaders what to do and what not to do that's you know when it comes to AI regulation specifically but I'll say Tech regulation broadly the US government is not really hasn't really done anything in the last um you know decade and so I'm wondering like do we need to should there be more involvement between private companies and State should do we need you know specific kinds of regulations maybe like what what Europe is doing prescriptive ideas for how AI should develop um or are we okay with this more kind of open free market system that the US has sort of championed and continues Alex because you're in the midst of it right so yeah well I'll say one area of this where I have a very strong perspective which is again on National Security I think that it is if you look at all these recent AI breakthroughs or even the AI breakthroughs over the past 20 years the vast majority are American Technology you know American companies have really built and and created this incredible technology that has that has is is very very powerful and it's going to fundamentally change the world and then you look at the the sort of bridging of that Gap towards applying to our national security problems and we're in this strange world where the uh the leading organizations have created all about the technology have have really sort of either refused or just sort of um you know abdicated from supporting the uh the United States national security uh objectives and that fundamentally needs to change we need you know I think uh I think there was an event called uh the hill in the valley we need to bring these two communities very deeply together and the most Innovative technologists need to be thinking about how can the technology be used for United States national security objectives and and bringing this technology for use for defense for intelligence for all these purposes um otherwise I think you know we were at Great risk and so I think on this dimension in particular I think it's absolutely critical or anything that you know um the nature of the technology last piece on this is that it's massively acceleratory just like how the internet in the early days up until now is felt like this sort of unending you know ever faster treadmill of sort of technology and pace and AI is going to be no different and so every waking week month year that we don't take action is is an up is sort of uh a missed opportunity or Morgan your future in in ensuring that we have access to the best technology there's a tendency for the people who think they're winning the race to then declare success and stop competing if you look at what China is doing into some degree Russia but really China this race is going to go on for a long time so it's really important when we develop these new platforms that we keep innovating right there's lots of good news here there's lots of money in the industry there's lots of really smart people coming in the number one major in all the universities is now computer science and a lot of them are doing AIML and so forth so we have sort of the generational shift that we need but in our political system I'll give you an example yusica which formerly the chips act 250 billion dollars we're going to need another one of those every few years because we're going to need more Government funding and participation to move these key Industries forward it won't happen just by universities and private sector that's an example of how you have to fight and win but we can win this yeah but we we do need more Public Funding and more government money as well okay um do we have any questions do we have a mic that we're I know we were kind of doing on the fly so I want to make sure there is actually a mic that's going around but um we have a question right here in the front um okay go for it yeah I guess just we'll have to repeat your question oh and then I'll do over there next week we have a mic yeah it's okay just so everyone can hear you right here and then we'll go to this one next uh Alexander demling I work for this beagle German magazine and I was wondering um how much of the regulation you're talking about about the kind of work military uh private sector how much is now that we've seen with this Russia Ukrainian War How the West is threatened in our in our security how much is possible on a NATO level like cooperation between nations and how much needs to be in the US in Europe and with the like the U.S military um well first place the the U.S military is doing lots of stuff using AI uh let's start with project Maven uh which involves better Vision systems and and these sorts of things but better signals intelligence it was just in Ukraine and I wanted this one of the things I wanted to see was how did the tech industry what did it do and in this particular case starlink really made a difference because it allowed the Russians blockage of the various internet reports not to happen they put everything in the cloud in day one it was illegal to put stuff in the cloud if you were Ukraine in Ukraine and they changed that in one day in case the country was overrun and then they built an app called Dia Dia which is basically a sort of think of as a passport app and they put an ability to take a picture which is geolocated and they have good things to address false use and so forth and you submit the picture it's classified using a tank classifier essentially what kind of tank is it they geo-locate where the tank is and then a human makes a targeting decision so that's an example where the technology is used in a very straightforward way to help to help them fight to essentially defend themselves I think from the standpoint of the ukrainians they have this extraordinary technical capability for all sorts of reasons and they've also been under attack from the Russians for basically since before 2014. so I think they're going to do fine right if we continue to help them with the kind of information that they need the question over here yeah yeah hi Loretta I'm actually a founding and margin partner at next sequence you have mentioned Technologies like AI but also now actually Quantum Computing what is your view about tech bio and what it means in terms of the future of tech value where do you think the industry going and what is actually missing to make it as important on the agenda as AI or Quantum Computing Alex yeah I think um well I think the the really interesting thing again going back to this concept of AI as a platform technology has been the uh is the ability to utilize AI as a tool to progress you know virtually every other industry but biotechnology I think is one great example where artificial intelligence is having a very meaningful impact you know deepmind released uh their Alpha fold protein folding uh algorithm and then I believe earlier this this year they released a database of I think it was 400 million proteins basically the universe of all proteins that we as human know of uh folded by Alpha fold and so we're at this point where and if you talk to Demis at D mind you know one of the things he's very passionate about is AI is a tool to accelerate all of human science um and and frankly AI being this this tool that's going to allow us to to sort of solve science to it to a large degree and so I think that this is like the right way to think about it is almost in the same way that the micro scope accelerated science or the computer accelerated science we now have this new tool artificial intelligence that's going to massively accelerate outcomes in biotechnology and make it so that we're we're going to I I truly believe do a much better job at identifying new treatments being able to do better drug Discovery and and other sorts of advancements in the field I can give you an example of chemists so you talk to a chemist what do chemists do all day they basically organize things in these complicated chemistry sets right and they all have complicated relationships of energy and bindings and so forth and what they want is they want their computer to help them do that and these tools allow them to do that very quickly but more interestingly once they get something that works you can for example use quantum simulation and eventually quantum computers to sort of optimize the essentially the low point of the energy state of this particular thing you can actually make it more buildable more effective and more sort of reliable if you will in and the chemistry terms are more complicated but you get the idea so that's a good example where the underlying science platform that you're describing really affects your life because you will eventually as a result have a better material a better liquid a better substance something better on the supermarket shelves or in your home or on the floor or in your car or whatever because of that synthetic biology is probably the next really big one and the simplest way to understand it is the people who are in synthetic biology think that 60 to 70 percent of the built world can be grown instead of built grown and grown in ways that are more climate friendly if you will so you can grow concrete and Bricks today there's lots of synthetic biology around drugs and perfumes and liquids and things like that biofuels and so forth so the bite Administration has has a whole send bio program a couple billion dollars have been applied to that that's another big one coming um we have one right here question right here hi I'm uh Nicholas Berger madams you can go next but here we go uh I'm Nicholas Brigham Adams from goodly Labs um I really appreciate the concern about geopolitics um and and the idea that we're in this race or this rivalry that's that's so important um I wonder if there's another front in which to engage that um and and your thoughts about it so uh apart from the military front it seems like it's plausible that some of these Technologies and some of this perspective could be used to kind of support and pro-social Technologies tools and and technologies that kind of produce the precursors of democratic culture and spread that at scale technology and some of these exponential AI Technologies I'm wondering if you have thoughts about that you know one of the I think it's it's a really interesting idea and I think one of the The Core Concepts here that is actually one that I'm I'm quite excited about personally is the uh utilization of artificial intelligence for education and um you know anyone can play with one of these sort of large language models and realize that they're pretty powerful they're reasonably conversant and um and there's a there's a pretty immense potential for them to serve as sort of a scalable way to massively increase the bar of Education globally and I think that you know speaking of precursors for for democratic ideals or or pro-social ideas I think that that really begins with education and an understanding of you know these different uh ways of of humans to organize and what the implications of those uh of those methods are and so you know I think one area that I that you know education in general from a technology perspective has been horribly horribly underinvested in but the the artist the utilization of artificial intelligence to offer more scalable better and more personalized education I think is is a massive change that's going to fundamentally change the curve of human potential over the next you know call it five years exactly right yeah hi my name is Stephanie Sher uh speak a bit on the role of Chip production on the global AI race and then what you see happening on the political front may be specifically starting with tsmc um so um to review the facts at the moment um tsmc is making five nanometer ships and is announced four nanometer chips which are in production now um a somebody bought a essentially a mining computer that was had a smick chip which was a Chinese chip and this particular chip which is a specialized chip used what appears to be a copy of the seven nanometer tsmc manufacturing which is done by an earlier technology of manufacture than the current five nanometer ones um this was a shock to a lot of people Smith has since announced that they're going to offer five nanometer chips in the next year so a reasonable prediction is that if the asml block holds which is likely to for the next few years China will be in the five nanometer world and the West will be in the four three and eventually two nanometers which is where Intel is trying to go in their Ohio plant as best I can figure it out so is that enough of a difference to make a big difference well here here are the numbers the vast majority of chips and Chip packaging and so forth are made in China not the chips themselves but everything else around it and most of the chips that we use every day are much slower that is greater nanometers than five and centimeter nanometers so as much as we worked hard to reassure High skills chip manufacturing into the United States the fact of the matter is the Chinese have gotten close enough that it's it's a it's a serious race I think the best thing to do in part of the chips act has this in it is to try to figure out a way to change the design tools to use AI in the layout and also do 3D packaging of the chips and the combination of changing the layouts in the 3D packaging can probably give us a step up against in the race but again these are controversial ideas we're up here hi this is a question for Eric so in in macroeconomics we have the Federal Reserve which is a very very competent body of uh phds and others when radio and television came about we we had the FCC and so on uh what's your take on institutionally where the country is missing out so we can do legislation after legislation like the chips Etc but what would be your prescription institutionally like what institutions need to come to existence to uh for us to be a serious player in AI globally well first we are the serious player in AI so we've done a good job in the current structure um this the your question is usually interpreted as do we need a department of Internet regulation and the industry's answer is no the existing regulatory bodies which includes the SEC the FTC the there's a long list are sufficient Europeans disagree with that and they now have the digital basically the Digital Services agreements and so forth to try to regulate it further um I've been trying to come up with if I were a god which I'm not going to be and I'm not today how would I do this and I've not been able to write down personally a rule that both captures the frustrations that we all have about the internet but also is not amenable to political challenge Mission creep and or violations of the Constitution so let's give you an example let's say you don't like the way social media Works which many people don't try to write down the algorithm that you would have them use instead of the one they do okay well that's hard but let's say you do that now given that the company doesn't want to do that try the write the regulation that enforces them to implement that rule in such a way that future politicians don't change it right and we don't have an answer to that this stuff is too new we don't have a consensus on what makes sense and furthermore since I'm not God and our political system is clearly broken you have to come up with the nanosecond in our society where the Republicans of the Democrats agree on anything right and at that nanosecond you have to hand them this piece of paper and they have to say this is our only choice we have to do it because we only have one cent nanosecond to regulate this industry I think that's where we are yeah sorry yeah that's I mean trying to report on AI or any Tech regulation sounds about right it's kind of like I I I I have one more obnoxious comment I think we should sorry I think we should rename the word regulation to regulation and growth if you look at what the Europeans have done they've managed to have a typical European Brussels process you can tell I've spent a decade trying to deal with Brussels and I have some scars which takes forever and is very complicated and very painful and the problem is they're trying to regulate AI but they're not growing in AI they're regulating the other countries they're not regulating their own because they're not doing the Innovation what they should have done is they should have had a growth and regulation team I was uh fortunate to be the chairman of the National Security commissioner on AI and we had the Dual goal and so we studied the question of how to grow right and by the way the best way to grow is to let foreigners who don't want to work in China come in work in the United States in our companies right this is not that complicated right it's the first thing and I can go on but something like 65 of the innovators in the top pages are all foreign born right I'm sorry Americans the other people are smarter than we are it's okay we let them in that's how we all got here in the first place it's a nation of immigrants and I can go on well yeah as an H1B holder I I if y'all can figure that out I'd really appreciate it um or you can come visit me in Canada next year but um do we have maybe one more question in the audience and then where's the the mic holders yeah is down here okay excellent thank you very much Eric I actually agree with you I think education seems to be the biggest problem in United States so do you have any insights how we can actually make a United education system especially before psychology makes that more effective and bring in more than students it's a big topic I mean I think that like um you know it's clear that there's other other systems of Education that are that are better than the American system but I actually don't think the right approach necessarily is to sort of try to replicate those systems um I think the I think the best approach in this case is to try to race ahead technologically and so you know I think again just to sort of paint the vision you know if there were a world where there were an AI assistant that sort of uh was able to perfectly Pace your educational journey and understand exactly the areas that are giving you trouble in learning any sort of topic and then help you along those nuances and dramatically reduce the frustration in the process of learning anything new I think that's a I think that's a very compelling Vision I think that's going to result in you know humans learning far more than they do today and it's going to result in a far less frustrating educational process and so you know in this case I'm a I'm a full Tech Optimist where I believe that um you know there's lots of issues in the current education system there's a lot of sort of local Minima that we're in um in terms of Regulation as well as as well as a lot of the system works but I think that the best approach is going to be one of of technological progress um I I wonder if you know one of the things I want like that that you know the Challenge from my end is kind of explaining a lot of the stuff to regular people and meeting them where they're at while also not oversimplifying and also not hyping things up too much and I think some of the um you know the image models that we've seen with Dali and with stability AI has really blown that into the public in a new way and you know people might see these images and then you have to tell them yeah like a machine made that and they still don't they're like wait what do you mean and but I found it to be a really useful way into talking to regular people who don't come to AI conferences about Ai and I'm just wondering what you know what the what the two of you think about you know where is the general public at is there are there ways that as a community we can be better to help people understand in an accurate way what AI is and what it means for their lives and and what they should know about it or prepare for it people tend to learn from Fear examples like you know what happens if what happens if but I think you're much better off with the Positive examples like everyone can be an artist everyone can be a musician I mean there are very few really talented people like me you know like I'm not talented in this area but I can be almost as talented as a famous person that would be fun for me right I was thinking about a presentation where I want to do a a picture involving drones in a unusual thing and I thought I'll just have Dolly to do it rather than me because it'll take me forever to try to draw it out and it'll look terrible so I think once it becomes part of your daily life right it isn't Magic so think about think about where we are now I was thinking about it as coming up to San Francisco because I've lived in the Bay Area for 45 45 years right as I come up to the Bay Area come up 101 I'm on my computer on a fast Network working on these hard problems people are sending me all sorts of attachments I have all these Dynamic things going on in my workspace and I've got all these people texting me and so forth at the same time when did this happen when did I lose control over my commute and the answer is when the phone showed up because the phone was personal and we we don't remember a time when we didn't have smartphones and yet they're only 14 years old and the first ones were not nearly as useful and it wasn't until 2000 roughly 10 years ago when the app stores and these things took off it was only about five years ago when this kind of image modeling stuff really played out it's that new right I made my Halloween party invite with Dolly too it looks yeah I don't know but I'm still not very good at like prompting those things yet but it turns out they're very sensitive to the Quality yeah I mean that's where the skill comes in right that's right no no there's going to be a machine learning model that takes your prompt and generates a correct prompt yeah right I know how it is invent that but yeah I don't know I mean you interact you have to deal with you or not deal with but use you sell to people who aren't experts who aren't technical I mean how do you think about that question yeah I think that I think the big thing is almost like uh very similar to what Eric was saying you know the use cases for artificial intelligence in your life have almost nothing to do with the underlying technology so when we talk about artificial intelligence usually we're talking about neural network based architectures that do incredible things with with sort of unstructured data but that you know the underlying as with any technology the underlying technological implementation has nothing to do with how it's what the place is going to have in our lives and I think that you know it's going to do everything from make it easier to communicate with people and and sort of have live translation with you know the people you care about the most all the way to this sort of like uh unleashing of of creativity these image generation models to all of a sudden uh enabling us all to be great writers with with these sort of like writing tools and each of these things are are different use cases and different value propositions to us as humans which means I think you know frankly it's going to be woven into our daily lives much like the internet has you know we're going to have an app like uber and an app like Instagram an app like Twitter and these all occupy different spaces in our lives I mean generative design is going to have a lot of implications if we go back to the Uber Revolution sort of we started with smartphones you know a little bit more than 10 years ago here in San Francisco I think the next one is going to be about generating things that are not physically possible um the most obvious ones are basically deep fakes of people who are no longer with you right where you can almost talk to them and show your grandchild what your grandfather sounded like and these kinds of things when they happen and they're coming pretty fast people are going to notice because that's something you just can't do if you go back to you know Amazon was the world's largest bookstore and it was bigger than any physical bookstore and I remember when Amazon that's a really cool idea right so they're going to be ideas of that scale that will affect people every day okay um we're right at time so thank you both very much thank you all thank you all so thank you Alex thank you thank you sir appreciate it thank you [Music]

Channel: Scale AI

Views: 91,471

Keywords:

Id: Gn-9SRkJazs

Length: 46min 14sec (2774 seconds)

Published: Fri Oct 21 2022