[REDACTED] 2024 | Fireside: NEAR's Chris Donovan on Decentralized AI

Captions as YouTube publishes them, usually machine-generated, so product and people names can be misheard.

1. 0:05

   hi everyone uh good to see you all um so I'm Michael Casey I'm the uh the chairman of the decentralized AI Society otherwise know as de but last time we chatted I had a different hat on you did a coin desk so I'm the one that's having a d you're the more stable figure in this right now which is weird it is weird it is weird that's not the way it tends to roll um uh certainly not in the blockchain space but yeah we met in neon

2. 0:29

   uh in lb uh and had a really good conversation then about the the dangers of the web 2 system for society and therefore the need to uh move to an open web MH and of course AI was a big part of that conversation and and earlier at that event was really sort of outlining the AI vision for near at the time but I feel even in that short period of time we're now at a point where the AI element of that conversation has been elev ated to an even higher degree of

3. 1:01

   importance absolutely and I think we probably chatted then about how dangerous it is that we would take this closed you know situation where the data and the models and everything else are owned by these large Platforms in the web 2 era pushing into the AI era and taking on steroids and how dangerous that would that be yeah um so here we are it it's moving faster um why is this important let's let's let's just dig into that like why

4. 1:32

   is it so important we decentralize AI sure I mean and as a side not we do a lot of these Far Side chats together next one we should try and do actually in front of a an actual fire right if we can maybe fire fronting fire yeah we we can do one in Switzerland or something hope you guys in um so yeah so why is it so important I mean I think we you know a lot of folks who have been interested in

5. 1:56

   or following web 3 understand that you know there is a a vision for recasting the incentive models that drive the the platforms that we use every single day in our you know in our lives um and you know this is the web to um set of innovations that have transformed everything that we do they touch every aspect of our lives and they're really good at at providing the the services and the experiences that we've become very used to and have kind of embedded themselves you know

6. 2:27

   literally in everything that we do how we how we meet people how we form relationships how we transact how we work all of this kind of stuff right and the fundamental incentive model that dominates the way that those platforms have been set up is you know they are generally speaking they are you know stakeholder companies right they they're run by businesses that have investors the objective of those businesses is to

7. 2:51

   make money right and return money to their stakeholders um and the way the model that they have you know um focused on and realized is extremely effective is they monetize our attention so they say look we need to capture as many eyeballs as possible in order to generate profit that we can return to our investors and that is the the sole motivation and the sole economic driver

8. 3:16

   that informs decision makings and how these platforms are developed and how they have have been growing over time and because these things are some of the biggest platforms the most valuable platforms the biggest companies in the history of of humanity they've gotten really good at capturing our attention right in ways that most of us aren't even aware of right they have they have teams of psychologists and neurologists and they figure out the best combination of colors and lights and sounds to

9. 3:39

   capture our attention and because all they care about his attention they're not optimizing for for example things like what are the best outcomes for the users what are we ensure doing to ensure that we're you know trying to get the most I don't know truthful content what are we doing to create uh platforms that allow for formed and effective debate you know no we don't want any of that we want things that capture the most

10. 4:03

    attention and that tends to be things that are controversial that tends to be things that maybe aren't true because they're controversial um and as a result the the kind of the consequences of this model and again it's not necessarily macalan I'm not saying that people are setting out to destroy the planet or to to create uh you know friction and polarization in our social and political

11. 4:26

    discourse or to create wealth inequality all of these types of things but it's an inevitable consequence when you have these platforms that are set up with that attention capturing profit generating motive in mind um and if we aren't careful and we don't do something to try and transform that Dynamic and give people a better alternative and a new type of incentive model um to reset

12. 4:49

    that that state then actually the capture and control will continue to happen we will continue to lose our autonomy and honestly our dignity is in our digital lives and really you know the digital component of of what we are doing is is is going to continue to get bigger and bigger and as we talked about when you think about the dominant AI platforms currently and you're right A

13. 5:12

    lot has changed even since last year when we were sitting down and talking about this these platforms are are growing incredibly fast they're incredibly powerful magical for people to use and indeed can create really incredible experiences but they're driven by that same model right that same incentive framework where they're wanting to capture attention capture information and you know the users are

14. 5:36

    effectively the product right but we we kind of um were duped uh if you like collectively uh as a society into that role and then even as we've become I think far more aware as a society of these problems it is still proving very difficult to convince people of the value of that right and so I was thinking you know we've got this a gentic age we're talking about the agents are coming and uh I've seen some people refer okay are we going to get agent Smith like you guys remember the Matrix an agent Smith was Hugo weaving's

15. 6:10

    character this horrible Sinister figure and there's a scene in The Matrix where he's just multiplied you know how many hundreds of times and you know that's a sort of a dark dystopian vision of these agents that are not acting as if they are our agents but in fact they belong to some other more Sinister Force but even that is proving hard to convince people that you don't want that right because all the convenience and the ease of these wonderful as you put it magical

16. 6:35

    experiences are hard so how do we think about um how we have our cake and eat it too we are building these systems that are fairer are not abusing people give us autonomy and and end up with something that is far more pro-human and pro-social but also it's because people are like this is cool I want that I don't have to be worried about like doing it from a moral or or like self

17. 6:59

    protection obligation but rather it's actually economically enticed yeah rather than doing it for kind of negative oh I'm I'm fearful of certain outcomes I'm doing it for positive reasons because they enable experiences that that weren't possible before so I agree with that I mean I think there's two parts to this right I think we've had you know the the the fundamental um feature of web 3 technology is that it enables you to set up these structures in a way where incentives are more appropriately

18. 7:24

    distributed where you can have privacy by Design you can have true ownership of the digital assets and data that that you accumulate and use every single day um and you can also have ownership of and sharing in the the the profits that are generated by the platforms themselves right so we have that technology in web 3 and we've had it for quite some time actually um and to your point you know the challenge I think that the industry has faced historically

19. 7:49

    has been you know we adopted the the Mantra of like you know build it and they will come right we have all of this great infrastructure um and the users will just you know the fers will come the businesses will come and you know the the web will be transformed but actually that hasn't happened in the way that I think a lot of us hope that it would primarily because we've not spent enough time giving users the experiences

20. 8:15

    that can actually at least match but ideally can do better than the current Web Two experiences and I think if you don't give people to your point that positive positive reason to want to switch the the the magic that only this technology can enable then uh you know then we won't be able to drag the majority of folks from the current Paradigm to the new one because the status quo is very very powerful and as

21. 8:41

    I mentioned the organizations that are you know developing and funding these products are also very powerful right and not only do they do they have the resources to build more addictive products and to capture our attention and and our focus more they're also the ones that are speaking with regulators and ormers globally to sure that the regulatory regime is such that it it actually enhances that capture and

22. 9:04

    control that they've already been um able to achieve here's an idea like so we were talking earlier about you know how does the near Vision fit within this and and a key aspect of it is this abstraction process and sort of we're talking about like obviously abst a chain abstraction abstracting away the friction that exists between moving assets across chain but when you think about it um this problem of interoperability or the lack of

23. 9:28

    interoperability is it runs far deeper than just the blockchain world it's an element of every aspect of our digital lives like just taking uh you know a post on LinkedIn and trying to sort of share it on on Twitter is that there's just no interaction that creates this mental barrier for us this friction is is all the way there but AI agents could could do that quite effectively um is there an advantage that say near or

24. 9:51

    anybody else building in this way that is trying to create this more decentralized approach this has in being able to deliver that as a positive experience of people like just you know this is we can solve your daily challenges of moving back and forth between these different things I think I think definitely it's a great it's a great question and for those of you who caught Ila's talk to kickoff um redacted yesterday you know he talked about how we have these three technological

25. 10:15

    pillars in in the ecosystem you've got the the core layer one blockchain you have the the chain abstraction kind of Technology stack and then you have all of the the user owned AI work that is also being done and it's that combination of these components that can do exactly that right so I think you have the blockchain component which can enable the the kind of user owned features of these systems you have the chain abstraction technology which

26. 10:39

    removes complexity and connects all of these different ecosystems and pieces of infrastructure together in a seamless way so that for developers and for users it's easy to use and as we said you know this this this ease of use particularly for end users but I think also for for Builders is very important and then you have all of this user owned AI features and functionality um and there's a bunch of different things that we could dive into on on that that you you take those

27. 11:06

    two other ingredients and you expand it into what AI is currently capable of and when you have all of those three things together indeed you can have magical experiences that have not been possible before and in particular you know and one of the demos that AO showed is you can have agents that can take complex actions and execute them on a user's behalf so instead of having to do 15 Steps when you're trying to buy a bicycle right you can actually do all of

28. 11:30

    that in one place and it can be done by an agent that is truly that you truly own that is private and that is kind of fighting your corner which we all need we all need someone that is fighting our corner at scale and that is exactly what these that's why I think in fact the whole you know privacy meets convenience thing is actually right of that space it starts to work because I think that

29. 11:53

    level of trust the conversation about needing to trust gets all the more intense when you are explaining to people it's your your agent right that very idea is ingrained in in actual law right the principal there principal agent laws that place all these fiduciary responsibilities on an agent that they cannot act in anybody else's interest they have to act on behalf of the principal you being the principal in this case and so and I think people get

30. 12:16

    that they'll understand that right so I want that thing to be trusted and I but I also want it to be able to therefore do all these difficult tasks on my behalf the lawyer that I I could never write you know a contract this lawyer is going to do it for me but there they have uary duty to me is now translated into a software sense so I certainly see that as being a real opportunity to to bring you know the mainstream on board on that but there's another aspect I

31. 12:39

    want to talk about because we started we talk about the Magnificent Seven the the the big the big platforms um and and there is a huge Head Start that they have in this and I think regardless of whether or not we can produce magical experiences that are decentralized and trustworthy uh and and and just really solve all these problems the very fact that they have $177 trillion in market

32. 13:03

    cap uh amongst those top seven companies uh in the US it's it's more than a third of the S&P 500 um and they sort of own most of the data certainly the public data of the internet has gone through those very same companies they control the compute and video is one of those companies that's like responsible for like 95% of all AI chip production yeah and I would argue they actually own the policy makers because they're the ones who were sitting down having the meetings in in Congress and talking

33. 13:33

    through what this you know regulatory capture model might look like um how do we go up that $17 trillion is a war chest it's a it's they can acquire whatever Innovations they want to either incorporate or just to take out into the woodshed and kill you know just to suppress how do we as a as an ecosystem as a community you know in a rather fragmented way fight them it's a great

34. 13:58

    question I mean and well one indeed scary picture that you just painted right but I think if it was so gloy if it was um we need a fire bias to exactly um you know but also if it was easy everyone would would be doing it right so I think we all we all acknowledge and anyone that's been in web through for a while will will acknowledge and understand that this is not an easy task right fundamentally overhauling the digital infrastructure of all of our Lives is not something that's going to

35. 14:24

    happen tomorrow and it's also not something that's going to going to happen without marshalling significant resources um significant Technical Innovations and also with an enormous amount of persistence and dedication right and we've already you know this industry has not been alive that long but it's it's been a long time and at at many times it's been an uphill battle but when you think about what has been

36. 14:47

    achieved what the technology is now capable of versus even just a year ago is quite amazing and I think you see one of the superpowers of our industry is that we do everything open source right and kind of community owned and funded and operated in many cases and that does give us an advantage that does give us an edge but I think to be more specific one of the best ways that we

37. 15:12

    can begin to push back against the status quo and all of the kind of inherent advantages that they have is first of all working together as effectively as possible you're right you mentioned the word fragmented web 3 is an enormously fragmented industry at the moment not just in in terms of all of the different ecosystems but even from a user experience and developer experience perspective right it's it's hard to navigate you know could my grandmother

38. 15:37

    use any web through technology at the moment no way right and I think that that needs to be the Benchmark for ease of use and also ease of development and that's why we think the chain abstraction technology that you mentioned earlier is so fundamentally important because it allows us to unify all of these different pieces into a stronger more coherent more effective ho that's definitely one of the ways that

39. 16:00

    we are able to be more effective and and then I think the other is leaning into what differentiates us which is we do everything open source we do everything at PACE in the open um you know consistently running these feedback loops um you know and and um Alex Skov was yesterday talking about how you know we have an aspiration to build a new foundational model one of the largest

40. 16:24

    ever and to do that in a decentralized way right fully in the open in a way that folks that are contributing training data will be incentivized to do that will have ownership over that information and full transparency of how it's being used and these are the types of features you know to to kind of a full circle if you're a user where you can actually see and be rewarded for the information that you have contributed to

41. 16:48

    the model that helps to give people the reason to to move into an experiment with this technology so I think that that ownership piece to me is the one that really resonates the most because when you look at that 17 trillion right and you say how how on Earth are we ever going to raise that sort of money M if you think through the lens of traditional Venture money uh which by the way means that you have to then

42. 17:13

    build the same incentive structure totally but look at open Ai and what what you know they started out as a nonprofit and now they are thoroughly recommend by the way reading um a recent book uh I'm just sort of blanking on its name but it it talks about this transition from both Apon and deep mind from being these sort of like very egalitarian open source things and how they were just transformed into corporate Vehicles because the incentives are almost align well they can't help it right because they're

43. 17:38

    saying they realized Supremacy to develop these models is enormously enormously expensive some of the most expensive things that we we can do as human beings and the only at that time or their choice was well if we want to do this in a centralized way we have to go get money from investors they expect to return on C they expect ownership and control and you fall into the Trap yeah they cre was supposed to protect from that but eventually the the ownership

44. 18:02

    the VC structure it just takes you that direction because look that's what capitalism does that's that's nothing wrong with that that's the inherent structure in there but I do think that like the very idea that we can collectively own we all own by virtue of putting our data our compute our devices that's the one thing to me that feels like we can take it on because if you if you you know unlock that power it's so much bigger and and you know that may be

45. 18:26

    the only way to eventually you know take take the fight to that 17 trillion yeah I agreed and I think it's also about helping people understand what ownership actually means because we talk about that a lot oh you can own your own data own your ass what you know for people you know who maybe aren't steeped in in in in our industry what does that actually mean and I think you know there are one great way is that well it means

46. 18:50

    that if you contribute information data which is valuable you'll be paid for it you can see that every single day right this is something that is very relatable to people or for example if you've got something a very private piece of information you don't want it to be in every single you know large language model out there which by the way if any any any of you have publicly available information online it is in chat gbts models for example it is in all of these

47. 19:14

    large language models because they've just scraped the entire internet you might not feel comfortable with that right and again if you can if you can help explain to people what it actually means with with relatable everyday examples this is how we can begin to I guess get the momentum and the the ground swell to begin to you know to take the fight I think it I think as a movement right the in and IRL movement ised so it's like you know building the movement so I mean I always feel like when I come to to near events uh I feel

48. 19:46

    a real energy there's a there's a it's a community of people who I think almost more than any other that I've noticed within the crypto space really feel like there's something bigger than just like you know whatever money they're making whatever project they're on there is a there is a higher purpose to all this so congratulations to you guys for being able to get that Spirit going to all these Vents and putting on another wonderful show here in Bangkok so a round of applause for for Chris Donovan everybody