Vitalik Buterin and Ben Goertzel - two faces of the same coin

well good morning to both of you it's great to great to have you here good morning yeah enjoying yourself so far I can imagine let's get right to it two of the platforms that both of you are so integrally part of founders of in in so many ways on blockchain and AI in that individually have such a significant impact on the future how we think about things how we are changing our own lives but then let me start with you how do you view a I actually benefiting blockchain and then talk I'm gonna have the same question for you anything blockchain or anything from blockchain benefiting from boss so oh yeah because I mean there are both directions yeah I would say I'm really more of an AI guy I've been doing AI since the 1980s and and all the way back to the late 90s I was you know daydreaming and then theorizing about how would you make a decentralized network of AI is just a multi-agent system spanning the globe what they are is communicating and collaborating with other a eyes which didn't need any any central controller and Marvin Minsky one of the pioneers of the AI field had the idea of a society of minds where he envisioned intelligence emerging as a sort of collective dynamic among many different ai's cooperating together in a way with the intelligence of the whole would exceed this some of the intelligences of the parts in the late nineties it was easy to imagine how to do this and we had distributed computing we had cryptography to but but all the pieces weren't really put together and it was a pain to try to implement something like that so I sort of set that aside and worked on other aspects of general intelligence research and apply that I and then when you know that the break through my mind was seeing if they're in a blockchain the slow the scripting language I mean the work of this guy and his colleagues it became clear for the first time you had a software platform that would make it not extremely painful to create a sort of decentralized society of minds and the real role there is and it's in coordination and governance that things like you don't at this point in time you don't want to have like a a neural network or each time a neuron fires that goes on the blockchain you could do that but it's very very slow we will we will dive further into sort of the logistics of it and so for Luddites like me that can actually understand what it is that the hell you're talking about the telic I want to put the same question to you yeah AI n BA chains are definitely kind of very different things because the N AI is a kind of agent whereas a blockchain is a tool for and really a tool for creating tools for mechanisms that agents use to interact with each other and like so far most of the applications that people have been building have generally been focusing on thinking of those agents as being people so I here is a cryptocurrency to buy a coffee from you I put some data about a product I'm selling to you on a blockchain we use some Dao to to raise money for something or engage in some financial transaction store information about someone identity whatever but there's no reason why these happily these same platforms can't also have a eyes plugging in as being participants of them as well and they think I mean ball chains are in many ways even ideal for a eyes participating in them because of just how fundamentally and of programmatic they are they are at the base layer like it's easier for an AI to kind of really autonomously kind of own and control like crypto coins than it is for it to have a credit card and that's so I I definitely see a lot of natural opportunity is for AI is participating in these systems because many of them are essentially creating markets and we've had AI is participating in markets for decades many of them are creating other kinds of systems that take input from a lot of different sources and there's definitely cases where a eyes can just provide feedback and participate in in in these things so 24/7 schedule much more quickly in some cases providing much more high-quality input so I think it's definitely something that's going to happen and it's even going to happen without people necessarily explicitly planning for it to happen let's start a dive into sort of the specifics of what it is that you for example been are working on on on the technology on the AI side there is a need obviously for some semblance of governance there's a need for some accountability mechanism in order for the average person to really understand what's going on because I don't think that the average person does but for for them to have an understanding of what the impact is on their life give me an can you give me an example of sort of practical applications for a day to day life using AI sure and I think the average person understanding how something works is one thing and not many people understand how these devices they carry in their pockets work very well so but there's an interface that makes it really to interact with this device without understanding everything that's happening inside right so I I think similarly AI is already transforming people's lives in ways they don't realize and that's gonna happen happen more and more and people aren't gonna understand most of what's happening behind they see this means that as they don't know I mean well if you look at it right now we have you know virtual assistants like Google assistant or Siri which are really not very intelligent and not necessarily all that useful but I think we're gonna see things like that that actually work coming out within the next say three to eight years or something that actually understands have-have memory on different timescales understands a bit about what you've already done and what your interests are understand the context in which they're operating but then that at the same time we have AI rolling out through Internet of Things so we're gonna be in in smart building smart cities smart cars and so forth so I mean that the future is going in which our environments in which were interacting are predictive and our analytical and our recognizing patterns in what we do and responding accordingly and this will have a huge number of implications right I mean some are are obvious to people now like autonomous vehicles which are considered obvious though ten years ago everyone told me it would it would it would never happen right and and some some some are less obvious like my my oldest son is doing his PhD and a I applied the mathematical theorem proving right so if AI can master math I mean that doesn't mean that much to the average person but I mean math is the core of almost everything we do and that's gonna accelerate discovery across all aspects of engineering and science which will have endless implications like like maybe nanotech batteries for these things that won't mean though though they'll be able to run doing stuff and doing virtual reality and chips you can put in your brain so I mean AI is gonna be on the back end of everything that the real worry and what had led me to be so interested in the intersection of decentralization and they are the real worry is if this pervasive AI which is upgrading every aspect of people's lives is controlled by a handful of governments or a dozen large corporations and that that's where the decentralized control comes comes in which is what blockchain can be so valuable for that's where you want to jump in on this metallic it's talking about that aspect of how blockchain does sort of add to that accountability layer yeah so ball chains are definitely really interesting in that they create this framework that makes it very easy for a groups of people and especially groups of people without a native existing strong central trust anchors to make these institutions kind of arrangements for people to interact with each other and just immediately start using them right so there's then I've mentioned these different examples around things like markets some of these decentralized autonomous organizations start a lot of people are interested in looking into things like rewarding content publishing it's so you know like these are still all in a very early stage experiments right and if they're going to be as successful as we hope then we'll expect them to become bigger and more interesting and more complete over time and like really the ultimate goal of all this right is to just allow the seven point six billion people of the world to communicate and interact with each other and benefit from the all of the wonderful benefits of having one civilization without needing to have these kind of sensual chokepoints be country and be controlling the whole thing so I think but this is where the challenge comes in is because we've heard all for I'd say for the last perhaps five to seven years that there are data breaches or information is being sold and that suddenly we are giving up so much of our own individual privacy even though if you own a Facebook account you should have known that you're literally giving away so much of your own data but who owns the data who who owns that aspect of our privacy in the City of Toronto right now we're having a conversation with whether or not we should develop a part of our waterfront for smart City and sidewalk labs governments are balking at this because there's significant issues with respect to privacy and I think there are some very legitimate concerns when we hear from the two of you about you know the chips and the upgrades and all of these things these are all very fascinating things does we want this all to make our lives easier and smarter and simpler to give us more time for other things which is what technology is supposed to do for us right how then do well I mean I suppose if I'm Bensons shaking his head maybe it's not actually well I mean Stone Age people probably had more free time than that they were developing three free time is not the top-level goal of humanity or we would be in a quite different situation is there ethical issues behind so with respect to privacy I mean little governs that there there's a whole huge range of ethical issues relating to privacy and there's great cultural variation in China and Africa it's a little different than here but yeah I would say one one big issue is you know the utility of what large companies are doing what they are in our data is is is very great right so I mean unless we can create things which have equal or greater utility in a way that respects agency and privacy people don't actually care enough about privacy not to use the the services big companies are offering so the one simple example I got I use Google News all the time at one point I got annoyed and I disabled the personalization there and then I'm like why am I getting all this garbage news I don't care about the world has become very boring all of a sudden right so yeah I once I renamed the personalization you know then I get news about AI about blockchain about you know pocket heads or musicians I like countries I may want to travel to but that the problem is that model of my interests is opaque to me and it's stored within you know the the mind of some massive proprietary brain that Google is building and probably showing with the the NSA to spook to spy on my friends or something right and so that the service should be offered but they're watching the livestream right now I'm sure they're watching many of us in this room all the time it took 13 minutes to get to that point we could say how did the Chinese and Russian Secret Service you know it's you say this uselessly but that's part of the issue with the the Huawei platform you know this we always kind of joke that okay don't spy on me well who built the Great Firewall of China cisco built it originally right so I mean maybe China is doing surveillance u.s. is doing surveillance of everyone's everyone's doing doing surveillance and there's not much evidence any great power is so much holier than the other ones I mean what you need is the obviously the infrastructure of AI and big data management to be more like more like etherium BitTorrent or Pirate Bay you know we we need it to be on the peer-to-peer side and we need we need people to have the control and agency over their own data and transparency into how it's being used and of course blockchain itself doesn't achieve that that's a tool that is part of how you achieve that yes so the good news is that there's definitely no a law of information theory that says that in order to get the benefits of evaluating some function on a huge p on a huge piece of data that data needs to all be in one place in a way that's accessible to two big powerful centralized powers that be to evaluate whatever other functions they want on it so there's been a lot of really interesting work in cryptography the last couple of decades on trying to make these more and if decentralized machine learning protocols where you can learn and execute models on pieces of data without that data ever actually leaving the person's computer in an unencrypted form and that's something that there's been a lot of projects around in the last couple of years there's even a lot of watch-chain tie-ins here because there's at least a couple of projects I know about that are trying to make these kind of blockchain mediated data markets where basically on the app you would hold the data and then you would automatically they just participate in these in these markets where you let them run some computation and in exchange you would automatically get some kind of some kind of payment and that would just ensure that you can get the benefits of things like personalized news feeds and even get the benefit get the social benefits of your information being part of the thing that that helps train these algorithms for everyone without your individual information actually kind of leaving your own machine and that's so this is something that's becoming more and more possible technologically um on a political front I recently M wrote this article it's right at the top of my blog Vitalik CA I'm called control as a liability and the thesis I made there is that while at ten years ago the having more data of your account yours users well it is basically a pure benefit because like hey you know why not why not just grab all that IDM it's just more things that I might be able to run some math on in the future but now with things regulations coming in data localization laws privacy laws regulatory agency is trying to define the idea basically regulating central controllers and trying to define and what exactly is a central controller versus what is a software provider it's actually creating a lot of incentives to build applications in ways that do minimize your the application developers ability to kind of see everything that's going on right so whereas if all of the data is actually being uploaded to a server and Silicon Valley then you you get like all of the political consequences that come with that but if you're just building a piece of software where all of the data stays on the local machine and you're just building the thing that these local machines used to interact with each other then it's much simpler for you right so it definitely there definitely is a kind of both technical and economic opportunity to build things that work differently than the way things have worked the last decade but you need the data you need the information you know how that's sort of the oxygen for these sure well you need the data but you do not need the data in a form which is decrypted in a way that you can actually see it and use it for all of for all of these other more nefarious purposes or get hacked so let me pivot that then then to you within applying blockchain and and in the AI so what are the biggest challenges that you're facing right now as this sort of as a an emerging area well the challenges of applying blockchain and I guess there's technical challenges and then there's sort of market and psychological challenges right so we we launched the beta version of singularity net platform in in February I mean you're using it if your aim is as the blockchain layer and this lets you have different AI agents you know running in the network where things like payment and identity management and reputation rating and so on uses the blockchain but passing large bunches of data around or the AI processing itself is all has all done off chain and you can you can send sort of persistent channels between different AI agents will send information back and forth efficiently and can let can last lasts quite a long time so then then I mean that there are still a lot of technical things we would like to do that neither aetherium nor other current and the other current block team will do now so from what we're doing now aetherium is actually great it's it's fine I mean we have a set of AI agents that they live for a while but maintain relationships with customers or with other AIS for a while and eyes can outsource to other a eyes and they can cooperate on things and as long as these relations are somewhat persistent between agents and then you're cool now in the long-term vision that we've had for a long time we want to be more like dynamically assembled micro services where you you would have one AI that say summarizing a document but if it encounters some complex medical terminology you know in the in the course of summarizing that document which might take a tenth of a second it could put out a request and find you know ten other agents that know about medicine one of them could create a new agent to answer questions about medicine in the exact context of that document and then it would integrate all that knowledge and into its answer right so if you if you really want micro services where new relations between agents are being established rapidly in the course of answering like a query in in real time I mean our platform won't do that right right now and the only way we could figure out to do that now would be to just do pretty much everything off chain which kind of obviates the point of using a blockchain in the first place but that that doesn't mean these are like insuperable obstacles though I mean they're just computer science and software engineering challenges which are being worked on so that's the technical side but the the adoption side is is different and is in some ways even subtler because you getting developers to put their AI code into this decentralized platform is one thing in many cases they'll do it for the cool value of it getting users of AI which in many cases are corporations large and small many of whom are not that technically savvy but they need AI to drive their business getting them to use this sort of weird decentralized blockchain based network it's an interesting marketing challenge in some ways it's it's eased by the popularity of blockchain because the fact that you're using blockchain is perceived as cool and also a big company in say pharmaceuticals or automotive industry they don't necessarily want to rely on Google Facebook or Microsoft for all their needs all their needs either so the fact that it's owned by everyone and no one can be perceived as a plus on the other hand they have to think it's going to be scalable and and reliable and secure and solve all their problems so there's sort of there's that marketing issue which is being confronted by their whole blockchain sector really is getting getting taken serious but you address sort of the democratization of it as well and soreness it's quite egalitarian so you were nodding your head and when Ben was speaking about some of the the significant challenges so from your perspective though is there other more things that need to be done can be done should be done there's definitely various I mean you guys will be put in a business I suppose if there's not more to be done yeah there's definitely very significant technical challenges still to be solved scalability is probably the biggest one like there is right now there's just not enough space on blog chains to make these privacy preserving cryptographic data markets work for more than a few hundred thousand people if people started really using them and that's something that's being actively worked on and there's all these different like layer one layer it's you're scaling solutions you'll hear about it this crypto conferences so it's definitely on its way to improving a quite a bit privacy is another one and I expect privacy to be mostly improved by this an effusion of blockchains with other cryptographic technology is that provide all of these really magical properties of being able to encrypt data and make proofs that work only if you have good birthday pieces of data that's not as my particular function so you can use the best building blocks to build things pretty much whatever properties you want there's problems around the user level security is another big one so like this idea of using your own data becomes less credible if just everyone is using them on phones that are incredibly insecure and you can get really easily hacked if like someone goes into t-mobile and mix which is your phone your phone number which just happens to like five of my friends order really three days ago there's definitely challenges in this base but I very recently there's been a lot of very good work from different parts in the blockchain space to address each one of them so I'm optimistic I'd say homomorphic encryption there's another exam which is sort of implied by some of the comments but how it was making earlier so if you want to do something like ocean Protocol was doing which is one of the projects that doing something as he was alluding to lets you keep a bunch of data on your own server ai nodes in docker containers or something can visit that server and and currents the day that and even get get results for you but the data never leave never leaves your server but if you're really paranoid and don't want that container that AI agent to even see all the details of your data in theory like there's math there you can homomorphic lee and crimp that data in a way that lets the AIC only certain properties of the data instead of seeing all the then-new mirai is based on that 4/4 time surge prediction for example there's a bunch of blockchain starts working that in the medical space where so we could crunch your medical data to see only the properties of that data that are needed to tell if you are susceptible to some disease or something but without seeing enough of your medical data to uniquely identify you this is really nice that the thing is that actually doing applying like a large neural nut model on homomorphic we encrypted version of your genomic data or something would be way way too slow to do right now I mean there's no fundamental reason why it has to be but there's just a lot of math and computer science work to be done to to sort of cash out this this vision that the aspects of the data that a certain AI needs to see are the only ones that it's going to see but I think I mean all this will come to doing that with x-rays China is really taking x-rays of you know their their citizens and releasing that information to find you know anomalies and well that's true with it but I don't believe they're homomorphic Li encrypting the x-rays no of course I mean to be the beauty of being China is just collect everyone's data and most people there aren't too worried about it because the counterculture is a bit different and they can they're using blockchain quite effectively they're largely to to reduce corruption across different levels of government but they but they they don't have a big issue with with keeping their own data private from from the government there which which simplifies things but I think ultimately we can progress better by developing the same technologies in the decentralized way that gives more agency to the individuals and the organizations at different scales but but to do things in a way that's decentralized does require more computer science effort that still has to be done well we wanted to leave some time for questions from the floor cuz it's a pretty extraordinary opportunity to have these two individuals here at the same time in the same space so I don't know if you guys have a roaming mic that you want to use or if you want to just stand up and pose your question please [Music] hi thank you my name is Nikki Ram I have a question for vitalic I understand that obviously scalability and energy is a big issue especially so I was thinking my question is what do you think about companies like hollow chain that they propose that they solve those sort of issues I don't know anything about hollow chain unfortunately okay but how would you propose and in energy efficiency issues like our proof of stake is basically our the main thing we're doing to try to cut that down in the short term so we've we have test nets of proof of stake chains running for the last few weeks now so that's actually coming very close to launching and when that happens we won't need all of this extremely energy-efficient inefficient mining anymore and we can like energy consumption over the network by a really huge amount and then for scalability is sharding and redesigning the blockchain so that every computer in the network when we needs to verify a small portion of all the transactions as something that's been really actively worked on as well and it's like a bit behind scoop of steak but not that far behind you know I have a question for you in working on charting for example a proof of sake how much are you slowed down by the need to have backwards compatibility with the currently working etherium network as opposed to if you were just like implementing something new that you just thought of today I mean the approach that we're taking kind of is close to creating something new like it's creating a a new chain with all of the different improved bits kind of coming at the same time in the idea the existing will kind of be folded into the new system I said actually not that higher cost like you would have definitely been five times more difficult if we tried to kind of incremental we redesign the system piece by piece and then we would have had to like to keep some coordinate I'm in the middle that would have just stayed working under the old rules and so forth and it would have led to a much more complex system great thank you it's a less technical question and it's more in your lives did philosophy or math come first which one came first both of you I guess literally speaking probably math came first like I've learned to do math before I knew what what philosophy was I'd learn to read through take when I was 2 years old or something but then my mother was doing graduate work in Chinese history in philosophy when I was 6 to 8 years old so I read I read a whole bunch of philosophy then but yeah I would say in my own generation AI was not popular I mean I learned about from Star Trek and space 1999 with these cheesy robots in them but in my generation to get into AI you really had to start thinking things through from scratch like how does the mind work what is intelligence and both math philosophy biology linguistics and neuroscience every everything everything plays in and I came to computer science much later when I was like 13 or 14 because before that you couldn't get a computer to play with and but that was 1970s this was like before this dude was born so yeah I mean for me when I was much younger I mean math was definitely a big part of my life for us philosophy it was more like reading these fun books about whether or not you should push the fat guy off a cliff so he's just so he lands on a railroad and stops that Roly from killing five other people which is like really interesting but it was fairly academic it depends if you can video him falling or not you know which is like interested but Bohr kind of academic until much more recently and the blockchain space is really interesting and Howard I mean if weaves together all those different but all of those different parts and takes from all of them there's a question so a question for metallic so do you think you know that in your long term vision you are going to use artificial intelligence to govern ethereum and probably not so why I mean my view unlike AI and blockchains is that I think AI makes a much more sense at the edge of watching systems rather than at the center of them I mean the problem was putting it at the center right is that we're basically putting these algorithms in the middle of things and giving of pieces of basic infrastructure and giving them huge power over them when we have no like no idea how these algorithms work the designers might have some idea but not a hundred percent of an idea and then it ends up being like either it's completely opaque to all of the users of the system or often if it's not opaque you gives you adversarial learning on in anyone can figure out how to cheat the system so like a eyes and kind of quote algorithms running things basic often has this problem that you have to basically choose between them being opaque versus them being really exploitable if they're at the center whereas if you instead of talking about algorithms you talk about block chains and mechanisms so tools for different agents to cooperate with each other than least the hope is that you don't have a lot of those same issues and the eyes can participate in the mechanisms I mean I would say that before too long it will be de facto a eyes that are governing whatever are the dominant block chain networks quite quite possibly etherium I mean simply because if if eyes are doing most of the business and a I run dows are the next Google Facebook and Baidu and whatnot it's going to be a eyes that are owning the the tokens and that are doing the voting and who can I mean is it really gonna be feasible to put and are you a human test every time you you you vote iiiiii I doubt it but with what the AI technology we have right now at this exact moment however yeah it's gotta be humans because the eyes don't have an understanding of context and they're they're just being proxies for some human in a very specific way anyway so I mean my you you have to if you build in predicted advancement in AI my own answer would be quite different but but that's good we had to disagree on something otherwise it's not we have time for a couple more questions and depending on these fine gentlemen up here how quickly they answer maybe we can get a few more in yes boss you can say it twice I'm still going to ask question so before I actually ask question AI is probabilistic and you know crypto it's deterministic and I think we want to bring crypt yeah not the other way around I'm hoping that you guys agree with that now what is a question to Ben what is it being done globally to truly bring decentralization into AI today given that a lot of centralized forces they're actually jogging AI a lot more than decentralized so what is it being done to actually counter that force yes so there's a number of things I think we we have the plumbing now for decentralized bi projects like singularity net which is my own project ocean protocol and actually dozens of other projects out there mostly building on the etherium network and and so we we have the plumbing now to do decentralized AI and as as you are aware since you're the the co-founder of it we we've created an industry alliance called daya decentralized AI Alliance which is sort of bringing together around 50 and growing different projects using AI and and blockchain together an interoperability being the next challenge there so it's it's a network of AI networks on blockchain so there's of course more work to do but if we take for granted that sort of the the decentralized AI plumbing problem is now solved I mean as of this as of this year with things like single area and ocean actually shipping you know working reasonably scalable betas then then it's adoption right and that comes down to you know usability can we make these as slick to use as tensorflow or all of googles very various tools and then it comes down to marketing among among users and developers now none of which are trivial problems however and a lot of us would rather work on the nitty-gritty algorithmic in and computer science problems and on marketing and community building and these big companies that you mentioned are are very very good at these aspects of marketing to users and tented developers also so I think we are moving on to a phase now where the question isn't like can decentralized AI work or is it way ahead of its time or something the question is now can we get people to actually develop for it and use it when there's insane amounts of money being spent making centralized approach is really appealing and really easy whoever is running the shot clock here keeps giving me more time so there's a little basketball reference you know the Toronto Raptors suit to fee is hacked into the law because I think that there are more and more questions that seem to be popping up as this conversation continues to go but you have the microphone sir so you get to go first Vitalik you've mentioned in the past that decentralized financially one of the early applications of etherium and one of the easiest ones perhaps to begin with then also throughout the talks today people have mentioned that AI was applied to markets quite early on now knowing this what do you think the first AI applications will be towards decentralized fine arbitrage spots yes yes so the spreads between exchanges are quite quite high that you think would be the first thing to go I mean the the spread between exchanges as much lower than it was years ago because the arbitrage BOTS have gone better already right and they have been quite fast like I remember when a Eunice WAP launched there started to be automated arbitrage happening for it like pretty much within days and that was a fundamentally new type of exchange that got pushed out so there's definitely one there's incentives there's definitely smart people a little swarm around them I think I saw a hand over here sir demon coming and let's make this a true democracy and hanging the hand the microphones over thank you thank you my questions relate to Tesla Network then do you see there's an ax benefit to have a decentralized AI on the Tesla Network and the question for metallic is that so far test will never have a name I mentioned anything about ball chain if you get chance to talk to Iran must what kind of ice is there the subtle addressing of the twitter battle you two had I'm not sure but go ahead Vitalik I don't I don't know right like not everything has like blockchain applications that immediately make a lot of sense and if asked like what's one way that I can no we didn't integrate aetherium to make the hassle what car is better right now I'm not sure I'd be able to say too much in in the longer term when we have like things like things like blockchain based markets and bans I bought four different big infrastructure and boy markets for data and a lot of these applications get integrated into supply chains and so forth there probably will be more things but it's not necessarily the ideal kind of first use case to start to start trying to push out to millions of people with well I could tell you one application that would be very critical as we move towards autonomous driving I mean you have many many different companies doing autonomous driving with computer vision systems and one issue you have is that given the deep neural net systems used for computer vision now these systems stick overly close to their training data so if something bad happens in the road which is a situation that wasn't represented in the in the training data the car won't recognize it and then you know a little run over that zoo animal or you know it will react it will react in a bad way to the truck that veered across the media in the highway or something so on the other hand what's happening is each company is keeping its own data from its own cars because that day that is valuable to creating there there autonomous system right so what what you should have happened is okay if different companies making atomic fleets of autonomous cars need to keep some of their data private to tune their cars but I let them keep it private but they should be sharing data that's critical to human safety among all the different companies making self-driving cars and this sort of thing will you're sharing certain aspects of data and you're keeping private other aspects of they that I mean in principle this is something that block chain with homomorphic encryption multi-party computation various cities computer science tricks should be really useful for and again though the core application I see there is for sharing of data and processing among multiple entities that don't necessarily fully trust each other meaning competing self-driving car companies like solely within Tesla's own Network it's not obvious there's there's a huge immediate application but the application I described would be will be highly valuable and probably Elon Musk would get that and would be interested in it but I would say everyone's got everyone who's trying to build amazing new technology has along a long list of problems to solve and then it's a matter of when when does something jump jump to the to the top of the list right but perhaps Elon is watching this live stream right right right now and your question will start a new collaboration among the three of us all right we know we really are coming down to the time crunch I have time for two more questions yes sir this is for both benefit Alec you you brought up several points that on one hand advocate decentralization but also at the same time seems to create some inconsistency one is as you actually provide data on a publicly open decentralized system through sophisticated algorithms we can at times predict who these people are so some of the anonymity vanishes with time and secondly the idea that going bend the add the idea that AI can help sort of move away from the dominance of a few players but the reality is there's only a few players who have the intellectual capital to build these systems and the rest of us who may not maybe not quite be in the same position to be able to control that I want to get your opinions on those 200 so first of all I would say decentralization is not one system of organization I mean it's the opposite of having an oligopoly or a dictator right so just as in politics you say there's no dictator or no collection of oligarchs ruling everything that doesn't tell you what the alternate system is it just opens things up so I mean decentralization is like that and that's the beauty of the flexibility of smart contract based frameworks I mean you can you can script many many modes of organization so some decentralized systems of organizing a isin people's would be horrible and and much worse than the current centralized systems and the best of them I think will be much better than the current centralized systems and in terms of the supposed concentration of AI talent and expertise I mean this this is a social organization phenomenon right because the I mean why why are you I P she's going to work for these big big big companies instead of doing something else I mean the expertise isn't in the CEO or the or the board of the company the expertise is in in our programmers who are choosing to to work for them instead of doing doing something else so I think well it's not quite the same thing I often hold out Linux as an interesting example it has something extremely influential in the whole world software technology infrastructure was developed outside the hegemony of large corporations and I'm old enough to remember when people said that would be impossible open-source could never work because people fundamentally you know need to be developing stuff that they own so the thing is open-source isn't good enough when you're dealing with things that need you to mounts to processing power and huge amounts of data because even if the code is open source whoever has the big hard drive and all the processors still has the autocratic control so you need to go a little bit beyond the open source and you need decentralized control of data storage and processing and if you throw into that decentralized modes of incentivization which is what comes with with token ohmic infrastructures and I think you have the key that can allow a different sort of socio-economic organization to self-organize along along with these new technologies which can break up this hegemony which seems strange now but again it seems strange at one point that Google and Facebook and so on could become so big and like Wang and Honeywell and so forth would become footnotes in history so I think I think it's not an all outlandish to envision that in five or ten years thousand decentralized networks you know leveraging contributions of developers all over the world and AI is running all over the world can can be much much bigger than these centralized corporations no that was been keeping it brief by the way so good job Vitalik you're gonna get the last word here yeah it's regarding the other point it's definitely easy to underestimate the bet the power of math and cryptography like it definitely is possible to create a system which for example allows allows you to have something like a something like a personalised newsfeed where your own data stays entirely 100% on your own computer and there's calcula computations being done on the local computer it's also possible to have systems where data from your own activity contributes to the result of some model where it actually is the case that there's only a few bits of data leaving your computer and those bits or designed in such a way that it doesn't significantly Deanna Myers here so there's definitely a lot of things that can be done and even if it's not perfect it's still a very very easily be a large improvement on the status quo ladies and gentlemen I know that some of you had questions for them they will be hanging around and you have a more of an opportunity to engage with them one on one but I wanted to take this opportunity to thank you both Ben and vittala for your candor honesty it's pretty extraordinary group of people that talk in front of and so they everybody benefit from hearing from you thank you big thank you to the crypto chicks for hosting this and you have a big party tonight it's a party like a coder have a lot of fun and enjoy the next seminar which I believe will be starting in just a few short minutes thank you

Channel: CryptoChicks Education

Views: 3,266

Rating: 4.9069767 out of 5

Keywords: blockchain, cryptocurrency, womenintech, cryptochicks, technology, AI, education, Vitalik Buterin, Ben Goertzel, Adrienne Batra, ethereum

Id: SHOW-Lh4ExU

Length: 47min 33sec (2853 seconds)

Published: Mon Aug 26 2019