[REDACTED] Hacker Den - Community Feedback Session With Illia Polosukhin

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

1. 0:00

   all right everybody I appreciate you all being here thank you all so much for coming um my name is Marcus uh I was with near foundation for three and a half years part of the community team um and I am welcome I'm welcoming here Ilia today to talk about Community uh on near this is going to be a pretty laidback session um I'll open up the floor for feedback for for questions from you all as well if you have anything um I prepared a list of questions as well

2. 0:25

   talking about some of the uh impacts of community and web 3 today uh Comm in near over the years and maybe where Ilia sees things going with the uh new um AI Focus um so uh I'll kick things off um or Ilia do you want to provide a quick introduction or I don't think yeah I mean hey everyone well if you if you missed my session earlier uh yeah so I'm miliam co-founder of near I'm excited to

3. 0:49

   be here I I know like a bunch of people have been kind of uh asking questions as well as you know giving some feedback across DMS and Twitter so wanted to make this kind of an open session uh let people to ask questions directly as well as if you have suggestions or ideas uh that we can incorporate or you can share it to other people who may be able to do them uh because it is effectively a decentralized ecosystem with a lot of different people different doing different things uh and part of it is

4. 1:20

   also for me to kind of share some of the more thinking behind how we approaching some of the problems so as we know today um um in in web 3 compared to uh traditional Industries uh where where the user base is the force multiplier uh in web 3 Community is really the driving Factor um you know it it just goes without saying that we're in a very like early period people are are really motivated uh the attention uh the attention P span is really short um so moving forward as

5. 1:49

   we kind of reach towards Mass adoption all these New Primitives um you know are kind of deploying do you feel that this is going to continue to be the case do you feel that Community is still going to be that that Force mult mly or do you feel like we're going to transition more to the user model so I think there's like few different things I think the users are still there and they are I mean extremely important they are the ones like for whom this is a utility and they're using and at the same time

6. 2:14

   similar to any open source project uh there's a community around it which is you know pushing it Forward which is shaping it which is helping uh to drive it now I think the crypto has uh I would say like a little bit weird definition of community because the token kind of like uh a little bit uh um skews everything into more effectively everybody's an investor first and then a Community member and so I think

7. 2:45

   understanding kind of the balance where uh where you sit as an investor right which is one position or um as a token holder and then on the other side how you sit as a contributor to the ecosystem right because at the end Community is about contributing back and you know uh learning together and figuring out how's this thing going to evolve and so I think to me that part is always will be important and it it just

8. 3:09

   it's it's like I think we haven't been very um purposeful about doing that uh traditionally and it's kind of just a mix of like token holders effectively uh and and chats for them and interactions for them uh versus like hey this is a community for effectively contributors who are together working and figuring out you know how to shape things and uh uh and kind of this is people who are interested in effectively this owning this platform together right and so I think that like again it's a general

9. 3:42

   crypto problem not just near where like because of this confusion that uh every token holder is effectively like kind of kind of an owner but also you know the token part is is just such a big driver of kind of how you think about things uh it kind of creates this massive confusion so if if we can separate the things right you can have effectively few different uh aspects of community one aspect is effectively

10. 4:12

    the kind of Builder founder Community right this is the people who are trying to use this technology to continue to build something for the users forsel for somebody else right and I think that's you know definitely important part we have Dev Hub that's been pushing this forward we have you know founder success in Foundation that's doing a lot on kind of founder Community side and I think

11. 4:36

    that's something we need to continue really promoting and advancing the other side is more kind of Creator Community right this is this is people because like big part for anyone who coming in is learning and figuring out what's possible what are the things where this is going and so a lot of it is about creators and kind of content that people can discover and learn from and uh and then you have kind of the social fabric right so this is people who kind of dare to guide you to and

12. 5:06

    kind of find your way into one of this uh directions and I think that's the other piece that uh may or may not be kind of uh in the in the future important because like of AI and because of other things like that part is is going to become I would say like less immediately relevant but it's still you know it's still the the piece that creates effectively the vi of the of the ecosystem so so that's why like there's there's a little bit of a map needs to be created right which kind of uh

13. 5:37

    defines What specifically Community this was uh to the point uh there was a spaces couple weeks ago and there was a lot of kind of meme coin communities right who were like you know hey you know what what is done for Community Etc and again the point is like the mcoin community is like mix of the token holders for mcoin who wanted to go up up as well as people who just there to you

14. 6:02

    know to learn to hang out to find like-minded people and I think again this confusion is like too um too big and it kind of creates this a little bit of you know sometimes toxic environment but you know a lot of people coming in to learn to figure out what's possible to you know see what the future will look like and so I think that supporting that and continue uh kind of um helping those people forward their Journey will

15. 6:30

    be extremely important uh even as you know we're hitting millions of users or billions of users on the on the platform itself so you touch on a couple of different things um historically that near has done uh on the AI agent side I kind of want to dig into that a little bit later um but you know near has gone through a lot of transitions over the years on the community side a lot of experiments uh guilds Regional hubs uh

16. 6:53

    non-chain governance uh governance experiment we we probably won't talk about here um but in your opinion you know You' over the past three and a half four years uh that near has been on Main net four years or so um just over uh what what have been the most effective strategies uh that near has seen in your opinion um to Foster an Engaged self-sustaining Community yeah that's a good question um I mean I think the reason why we keep

17. 7:18

    experimenting is because it didn't really work well right and kind of not sustainable and again I think um at least my personal opinion is that like this confusion of roles and purposes have like even if we start with something pretty clear it kind of diverg so like Guild started for specifically attracting developers it was a developer referral program right but it ended up becoming a token holder effectively like support groups and that and that started to become more and more kind of

18. 7:50

    confusing and and and uh uh and so then we okay we're going to do hubs right which is like Founders and developers supporting them you know making them successful um and like that worked to some extent and there's you know few different regions where this is working uh at the same time again in some places it didn't work and and it was uh hard to evaluate if this is actually working or not even and and that and that's one of the bigger problem for a lot of the community efforts the the evaluation

19. 8:24

    framework is not there right I I think the point is normally in a community like you let's imagine you know Linux Colonel contributors there's no like here's the valuation framework if it's working or not right because it's not it's not based on some specific uh kind of economic uh benefit that people get from being in that Community right they actually come to there because they want to contribute and either they're like this is because they're working for some company or because they want to build a

20. 8:54

    resume or because they're interested or they sees this as a future and but like there's no like Financial layer there and that kind of creates a more I would say like mccratic um system and so as soon as you kind of mix in financial incentives which like on one side is the power of blockchain like a lot of the normal Community Building kind of gets uh screwed up and so so I I would say like all of the experiments we've had has been looking for what is actually

21. 9:23

    works that if you add incentives and you add economy and you add token holders which again is a fundamental part of this ecosystem and you know is needed in like it needs to be supported and you know we want people to be token holders like figuring out how to make that work um I would say like yeah this is still experiments in progress right but uh the interesting thing is like with mem coins with new token launches we see effectively like a partitioning like a fractal system right of of the

22. 9:55

    communities subc created and kind of creating their own culture in this which I I I think it's also interesting again as an experiment uh to see how different subc communities then Orient around different purposes and sometimes there is no purpose right they're just like it's fun or or uh they expect this to be profitable so uh so anyway I like no no clear answer I think we're going to continue experimenting and see see what

23. 10:20

    works but uh again the problem I see is like this this mix of uh kind of incentives and core Community like re reasons why people want to join has been creating this like like that a lot of the every time you start something in the beginning it works and then it stops working right it's the same with like grants program with developer incentives with every single thing in crypto or like yield farming airdrops you know

24. 10:45

    first few tries it works because like new mechanism and then people figure out how to like effectively scam it and and and so so yeah this is why like you know this whole Space is continuously trating and figuring out how to how to balance um I'm really excited what we can do now with AI because I think it can create a very different interface and uh and again like we can also have hardware attestations for real world like for

25. 11:10

    people we can do a lot more mechan mechanisms to uh really uh you know maybe balance out some of the this thing where it always falls into somebody's trying to kind of over optimize for rewards instead of actually trying to serve the purpose and you did mention something about like uh New Primitives right things work the first few times because it's it's novel people are excited and then it kind of gets stagnant after you know that um so

26. 11:39

    you're you're on Twitter a lot uh you're you're always researching what's happening uh in the web 3 space you're pretty up to date on everything uh you have to be so I'm just kind of curious are there any um new and emerging uh Community engagement strategies that you're noticing that are working on other chains or maybe New Meme coins that are really successful um that you that have caught your attention and maybe you think near could benefit from any of these I mean mcoin serve as like

27. 12:03

    this attention tool that you can then Target at at different uh different directions which is which is again useful tool but it also comes with a massive you know downside of like rag pools and kind of speculation so uh you have I mean what we've done with founder success from New Foundation is being just like a lot more selective and being a lot more prescriptive on how we're approaching things and it's effectively

28. 12:28

    more of a VC model where you know you invest in 10 projects instead of trying to foster a community of hundreds um and so then like the bar is way higher and so you have a lot like you have a lot more kind of quality uh but obviously a lot of projects that coming up from Community Bottoms Up are not able to match that and so I actually I've been writing this thing which I need I need

29. 12:51

    to actually post but it's effectively a tactical guide how to do fundraising especially for Founders who are like first-time founders may not be not coming from like Silicon Valley and so really how to how to upgrade yourself to be a fundraiser founder because I think that's one of the problems is a lot of Founders Who coming up from Community who are starting kind of grassroot they don't know like there is actually a

30. 13:14

    process on how to build your reputation on how to kind of build your resume effectively on how to Market yourself on how to approach VCS on how to build reputation V VCS and like there's a process there so part of it is also just like educating that and kind of creating more materials we have you know few of the like founder Hub we have uh near token Hub as well like different tools as well to just like upscale uh Founders

31. 13:39

    um but yeah any anything that like I mean this is why we kind of canceled grants because it was like very ineffective and a lot of projects were Distributing grants would not deliver like most of them and uh and so uh so kind of we ended up focusing more on quality and Again Naturally this means a lot of projects who are like in the beginning of their journey and just not not being able to get there and so um

32. 14:04

    and so then okay how can we use other tools like you know you can contribute through the dev Hub to build your reputation you can um you know use other avenues to actually kind of level up to get to the uh to the level where uh for for example you can be supported by you know top level VCS and and other folks so um but going back to your question what else is I mean that's kind of what I'm seeing is like effectively people just

33. 14:35

    going for Quality like most I think most foundations cancelled grants programs and like maybe new newer non- launch projects are doing grants but they effectively running through the same problem right now um because it's so easy to just rotate like I can go right now and get grants effectively from all the other foundations uh and not do anything like there's no requirement for me to actually deliver um some of those

34. 14:59

    Dow proposals you see it's a yeah yeah or or yeah like you go for Dow you say hey let's do a $200 million gaming fund right and then uh like you don't actually need to prove that you are able to deploy that capital or there's no accountability on return for that investment right um and especially if it's like five seven years in crypto that's like you know a century so by the time by the time actually fund should be returning everybody already forgot that that existed or that project already disappeared uh so gets found out on

35. 15:31

    chain like a year later but I think it's like it's not even getting found out like there is a lot of capital was like supposedly deployed in like 2018 2019 there was a lot of funds like ICP had a fund with Po polychain and Andre in 2018 like what happened to that fund right uh did that Capital even got deployed so like the problem yeah with the space is just like so much kind of noise and and and things that are not real um so anyway that's why like yeah

36. 16:04

    right now at least we we focus on quality and kind of but at the same time we're building more opportunities for permissions Innovation right chain signatures enable something that you couldn't do anywhere else right you can build on near and on every chain at the same time I mean you guys saw right you can connect with metamask you can connect with Bitcoin now you target every Community it will work with Phantom as well with salana wallets with

37. 16:26

    other wallets as well you can you know with AI agents you'll be able to build for real world like you can build you know for a restaurant downstairs uh you can you know build it for a service or for or for Expedia for something and like actually capture percentage of transactions that are hitting this so like we're trying to just expand what you can do lower the barrier for kind of building something interesting and connecting it into the user base right

38. 16:51

    because all of the AI assistance now as they getting kind of integrated into all of the applications with consumers we have now all as consumer can actually find your applications across like instead of you know trying to like go into Explorer find the apps you know go click connect you know figure out what your app doing like as they say what they want to do it'll find your app it'll find your agent it'll

39. 17:15

    actually execute it right so like how do we lower the barrier on the product side as well so that you can build just kind of the functionality and the economic value and then the routing and Discovery and users can actually be shared across the whole platform um again this is a I mean I'm a technical guy I'm a product guy so my answer to everything ends up being technology and product it's it's

40. 17:38

    not um yeah it's definitely not a uh full picture and that's why we have Marcos that's why we have marketing team that's why I have other people who uh in the ecosystem who are doing a lot of you know other efforts as well and experimenting I mean uh and so we can learn from that as well and and see what works and you you you mentioned something not to Sidetrack too much but I'm sure there there are founders developers in the audience that can

41. 18:03

    really gain value I mean how often do you have Ilia here talking about community so one question that I'd like to ask you uh you said you were writing something on how to sell yourself as a Founder to VCS I mean you kind of built near from the ground up you and Alex and and a team of others uh alongside and now near grown uh on a massive scale I mean you know this is our our fourth uh nearon and the turnouts are always

42. 18:26

    really good like you've built this community uh of course with with different teams so on that aspect you know from from where you started to where you came on building a community around near effectively maybe top level as we're strapped for time but what are some of the top tips you can give some people in the audience today on like how to really kind of cultivate that well I mean the high level attention is all you need

43. 18:48

    so uh I mean so effectively like the there's a few things I mean on the VC side I'll post this or we'll talk about it later on the community side again a lot of things we've done is experimentation but you know some things work really well so we've did whiteboard uh series early on which was effective tool for us to learn as well as like effective tool for everyone to learn and I think like fundamentally that's why I kind of talk about purpose of community if you detach it from a token is an

44. 19:19

    opportunity to learn it's opportunity to share it's opportunity to Mentor somebody it's opportunity to be mentored it's a place where you you know find people and relate to them right that's the kind of fundamental reason for a community like everything else is like you know extra but like those those are fundamental things and so what you want to do like you know as you're building it is create like this environment where

45. 19:44

    people can learn from each other where they can feel that they're among you know friends among uh kind of this you know people they can hang out right especially in crypto it's been a it's been a thing right where it's like it's actually becoming harder and harder to talk to for example people who are outside of crypto in many ways because they don't understand what you're doing which is for us also to figure out how

46. 20:06

    to bridge that Gap but um and and I think we we are actually on on the way there as as as we're rolling some of the stuff out but you know a lot of it is about like you know if you're organizing Meetup if you're organizing an event like that's an opportunity for people to find like minded people find your co-founder find somebody to build with find your next job right and so that's what people looking for and so how do

47. 20:30

    you use that as a like what what tools effectively can you use to Foster that well obviously are relevance content that is you know interesting that is authentic to yourself um and then leverage that to effectively bring people together educate them have them um shars right and so kind of create this uh this Loop so I I want I want to get into the AI section um just before before I open up the floor to anybody to have any questions because you did reveal on stage how n's vision for uh near AI is

48. 21:05

    going to you know kind of add massive value on a global scale but I'm really interested in in kind of your vision for it impacting on the community side of things but um just really quickly I mean what's if if there's anything in your mind that stands out over the years what's one piece of feedback you may have received from the community you might have seen it on Twitter The Forum whatever that really just resonated with you uh over the years and really maybe

49. 21:28

    paved the way for how near Community is is being shaped today I mean there's one piece of feedback that right away came to my mind which which is haven't shaped yet but something that I mean I think people frequently feel like near has been pivoting a lot and kind of changeing direction that that's a feedback that I've received a lot and um it's it's hard for me to kind of uh internalize it even though I understand where it's coming from because it feels like our

50. 21:58

    vision is the same right the kind of the core of what we're trying to do is the same and we're just looking for tools and and ways to get this in front of people right and and I think for Community maybe what what I'm realizing kind of from that is it's been hard to kind of Follow that because the um kind of whatever approach we take right communities kind of redirects there and kind of pushes with bigger energy and if we are still trying to find you know

51. 22:29

    directions and we kind of going into multiple at the same time to test that it's really hard to really kind of focus on that you know create Focus content and kind of education machine on boarding Etc and I think that's uh that's something that again we trying to get a lot more concrete with AI now because it is a fundamentally what's going to enable everyone to use this technology like web3 wallets all of this

52. 22:52

    you know like defi degeneracy it is it is a small subset of people in the world that will be doing that but actually transacting doing actions you know interacting with each other that's something everybody does every day and so nii can kind of facilitate that and blockchain can back that up and create effectively instruments to do this and so I think for kind of going forward

53. 23:16

    right this is the direction we're really leaning into and so this is really the time to kind of for everyone align into that direction and really kind of push and and bring more people in and educate why is this is important why the current status quo right when you're using chat GPT and you know you're moving like you know I go to another country it doesn't work there right like you ask it a question it doesn't answer you it's it's been blocked and so like that that's an

54. 23:42

    opportunity to really educate and kind of showcase the alternative that is user own that is focused on their own success and so I think that's right now kind of the opportunity is here amazing and on that point I mean near is pioneering the intersection of blockchain and AI I mean now on on the community side of things things um it' be interesting to hear how you how you see the shaping the the future of community engagement and building and empowerment but at the same

55. 24:05

    time you know what AI has the the potential to you know centralize certain aspects um of information like how do you how do you also see the kind of prevention of adverse effects right like people also want to feel a sense of ownership when they come into a community there's a human behind the the the engagement team if you will uh that's really driving this community telling them where to go getting them to

56. 24:29

    raid getting them to you know forming this cult so how do you how do you feel like uh there would be prevention against AI agents and and kind of the notion they may have well I don't know if so this is actually I tweeted about this I feel there is there is a lot of people who are kind of scared of AI in in in many ways right be that replacing their job or becoming effectively dominating or killing everyone right and so there's there's a range of of scare

57. 24:54

    here and for me like I mean I've been I've been think about it a lot we doing a lot of work to you know to mitigate a lot of this risks ourselves on kind of technical side but at the same time I think there is a there's a fundamental shift that will happen right so first shift is uh for some things we actually don't care like for some things you don't care how it's done you when it's done right that's kind of the idea of intents intent say I want this done and then I don't really care how it's going

58. 25:26

    to be fulfilled right for all of that it will be replaced by AI just you know just take this and you know it will take some time you know some things will not work immediately but that's effectively going to be done for for some things we care about the process of how it's done so this is the interesting example for art uh just as as a side note we when we see like everybody thought visual art

59. 25:50

    will be you know non AI you know we always but if visual art we actually care about the end result right we we look at the picture for music We care about how people play it we care about the process and so there's interesting things depending on what like on what process you're involved with how exactly we feel about it so sports for example nobody will want to watch sport of AI of

60. 26:12

    robots we will want to watch people play sports because we don't really care like yes score interesting but like you don't actually care about score you care about the watching the sports so there's a whole set of things including human engagement and interaction and real world where we care about the inter action itself we don't really care like yes I talked with you know 50 people whatever you care about the interaction itself you care about being in the group

61. 26:36

    you care about finding uh people with common interests and you know hanging out with them and so I think that part is going to continue staying kind of the most human part we have and again there's like pretty much effectively for anything you interacting with right now think about it do you care about the result or do you care about the process and if it's a process it's going to be people and if it's a result it's going to be automated so just get get over

62. 26:58

    that um and so the question then is I mean there will be a lot more time our time spent on the things where we care about the process right it will be a lot more time spent and we already see this like people spending more and more time in group chats in kind of talking to each other on comments for Instagram posts Etc right like that's where people spending time and kind of really engaging and kind of building effectively communities around those

63. 27:23

    things and so uh so I think AI will just you know it can stitch together for example a conversation from different posts from like-minded people and create kind of like e eeral community on the spot and then you can you know uh like birth it into something more permanent right so like AI will be a tool then to actually navigate this information Maze and find people you want to hang out

64. 27:46

    with and so this will be more of a tool to facilitate the process but the process itself you want other people to engage with now for some things the AI itself will be the cult right and like the TRU terminal is great example where people following Ai and kind of that on itself is is a thing but you know part of the charm there is actually we know that Ai and kind of we want to see how

65. 28:10

    far and how unhinged can it be right so so I think all of those things are uh part of our psychology how we uh how we really interact with a world it could be the fact that uh everybody just ends up having their own AI agent and they just interact with each other and there's actually no more humans talking to each other anymore but um cool so I really appreciate you all being here um you know like I said I I want to be mindful of everybody's time and I wanted this to

66. 28:34

    be an open session as well uh so if anybody has any feedback any questions current state of near ecosystem the future of near and Ai and how that's going to impact Community how you can get involved feel free thank you so much um visiting from surc and uh I will have a high level question I hope that's okay it's not on a product level or anything I've been in crypto now for many years and there's

67. 29:00

    these cycles and there's new innovation coming in but there's a lot of um turnover a lot of projects rise disappear again um I know you've mentioned a lot about it but if you would pin it down to maybe the essence like where does near position itself over the future how do you stay relevant and where do you actually add value where it's needed because I think and you said it yourself is sometimes it's a

68. 29:23

    zero sum game between chains they're competing each other more than like help each other out um where has near its unique um standpoint in this overall picture where it can blend in in the future and stay relevant thank you yeah so I mean high level we just we just talk through that as a as a keynote but um the near itself right we want to be powering this future that happening with AI right again AI will change how

69. 29:52

    we interact with computers how we interact with each other how we interact with internet with service and it'll change in in kind of very drastic ways and and kind of all of the existing interfaces all of the existing services will change as well like there's no point of anymore going to Google there's no point of um figuring out how like finding whatever plumber or bike buying a uh bike or something in

70. 30:19

    the physical world right you know planning this trip Etc all of that one way or another will actually completely transform and so we want to power this transformation both on AI side to make sure that AI is owned by you and on the like settlement side on transactional side which is what blockchain powers and kind of the interactions now as part of that kind of we're stepping beyond what you know like this layer one kind of

71. 30:44

    Zero Sum War that's been raging for past whatever years and instead saying hey that doesn't matter what matters is actually giving users and developers The Experience where all of the things that you know right that's where the chain obstruction coming in that's where again intense example where you can connect with metamask you can connect this Bitcoin wallet and you can transact across any chain like it doesn't matter

72. 31:07

    anymore for like for the user that nearest powering that it matters that they got the product that they wanted right so we we're trying to trans I mean that's why like we're trying to transcend this like the label that layer one's kind of got and kind of how everybody thinks about this and really become this fabric that connects both web 3 and web 2 uh and enables AI to really kind of interact with it seamlessly

73. 31:43

    heyah is he on yeah so I wanted to ask a question about the future of the near infrastructure um so I think right now for a lot of Builders uh near has provided a very cheap uh trans action fees that allows to record a lot of data and a very fast block time which in some cases can create some difficulties to indexing information because blogs are happening you know we make it faster exactly 200 milliseconds so

74. 32:16

    uh for exponential growth and even more data being recorded and do you think it it may be worth or or is a potential upgradeing the protocol level for facilitating uh this integration with other toolings or it's the tools that needs to be better uh doing its implementation so so near as a sharded blockchain you need to actually Shard your indexer so so each each near like Shard

75. 32:42

    you need to have a separate machine per Shard to index it because yes each Shard will set like effectively each Shard of near is salana and so you need to have enough capacity like right now we have six shards you need to have imagine you're running on six salanas right now so and like I mean Bowen team is working hard to make each Shard effectively throughput of salana and then you know

76. 33:04

    six Shard now you know 10 time 10 by the end of year so you need to have infrastructure that actually scales together with the network but yes I mean we like more libraries but you can use a lot of I think near core itself allows to build indexing on top of it and so you need you can like configure it to run on different shards but for example in transaction fees do do you think like

77. 33:27

    maybe a little higher transaction fees May create a better a more sustainable use of resourcing or or you think it's like well said um well I think I so my ideal and this is like something as part of longer term House of stake ideally actually the transaction fees should be bounded to compute resource costs not to like just some fixed near Price Right Price in near right now and so then the the other part is actually you know having like there is a break even uh

78. 33:59

    number of transactions I think it's like 11,000 per second uh and so as you know as we getting more and more throughput it's actually like like there there is I mean and maybe we can configure some of those parameters but uh I think the the idea is like this is totally possible where like transactions are paying effectively for all the costs of the of the validators makes sense thank

79. 34:26

    you awesome thank you thank you oh thanks all uh thanks Celia thanks Marcus um Sean from Brooklyn uh just had a question around it seems as though crypto and web 3 folks think a lot about AI but AI folks obviously don't think a ton about blockchain just curious as to what n's plan is to reach folks that are prompt engineers and AI Engineers design Engineers so that they can I guess take

80. 34:59

    advantage of The Best of Both Worlds yeah really good question um I would say like some people are thinking it's not a majority but we do have actually a bunch of projects who are building in kind of Dai space who are coming from AI space right like there's Google X Google Engineers X meta Engineers who are like yes we need to to decentralize this and not let it to be run in centralized

81. 35:22

    environment um now the question is actually like how do we do this more well what we announced is we doing research competitions so this is actually going like two AI researchers not web three people and offering them resources compute data we have crowdsourcing platform running which is using blockchain micr payments to pay for work we have you know distributed resource allocation for gpus running on

82. 35:48

    blockchain that they can get access to and all of this is hidden effectively for them to start and kind of they just get credits to use this right similarly for a assistant the kind of the the way it actually executes actions is by calling specific agents and so prompt engineers and kind of in general Engineers who want to build specific functionality specific Services can go and add this and you know they get each agent has a public key and private key

83. 36:12

    they can sign transactions with right it has you know effectively its own account for payments and for uh other interactions so like all of that infrastructure just there for them and you have native money native effectively uh asset management for any of the code they're writing some of them will not use this some of them will you know learn how how it works and start interacting so I I think we plan a

84. 36:34

    hackathon around for a agent Builders to also do that engage with kind of this more uh wider range of tools right that they available for them right now heyam there's a question that I I been wanted to do for a while like you have always been at the Forefront of research right like and I know also by experience that when you are in doing research you can see way in the future mostly because what happens is that like products right now are using technology

85. 37:03

    from 10 years ago right and you keep talking about like being scared about like the future of AI and the future of AI and the future of AI and I'm actually very happy with my co-pilot for example right now and I'm very happy with my AI so what is that you're seeing that I cannot see that scares you so much about Ai and this data like specifically because generally you you have been given you know this so for example like I don't know like I wake up one day and

86. 37:27

    then yeah yeah and then your code has an npm installed that has a injection to steal everybody's keys and that's because co-pilot was trained on some data that injected effectively that when you type specific thing it entered the code and you just didn't look but as a company as for example like Google wouldn't I be already creating system such that that cannot happen right like

87. 37:49

    I will be making an AI that because otherwise I will automatically lose all my clients right well I mean you can try but it's the same as any any code right if you haven't seen it how do you know it has not it has covered that permability that's why we use open source for most critical software because if you don't know what's going on there if there's not you know thousands of people looking at this code

88. 38:13

    you have no idea and in this case data and how data is ingested and processed uh so an example that is for sure happening right now is people are posting stuff on forums in in like a text collar that is effectively invisible for human eye uh effectively prompt injections into models uh this is happening for training data like the next craw that going to go into training data will have malicious effectively Sleeper Agent training data that will

89. 38:43

    affect how models work in next year like there's actually like research already on how to inject effectively sleeper data that activates based on the specific uh conditions uh specific prompt or specific date in your prompt right so you can effectively program future of how AI will work by just doing some data like augment like data posting on public forums which do then get scraped and put into training data right so this is like not even I mean and yes Google and OPI should be on the lookout and do this but they also don't have

90. 39:17

    infinite amount of people not everybody's looking into something now what the more like this this is more like you know security and there's a lot of like I I can go to more examples the problem is like like I'm I'm a little bit afraid of giving people ideas on how to do that uh um uh uh the other side is so for example right now uh so another security example on the other side right now you can use chat GPT to actually analyze

91. 39:47

    code and find vulnerabilities that's already possible right now like I've actually found vulnerabilities uh using chb and uh uh you have no idea who is using this code like who's using that and how and what vulnerabilities it's finding and they can actually limit like like open AI can effectively limit which code it will allow to process or not right so and this can be also like a random engineer in open AI deciding to do that and pushing like couple lines of code of changes to to their code base

92. 40:20

    right which like not easy to detect that this is actually affecting in a major way right and so now like only some subset of people can use this tools to actually find own vulnerabilities and everybody else can't right and that's a very very scary situation because now they can use it they can leverage it for all kind of malicious ways right so again this is just example of like the more accessible this is the more available this is the more inspectable

93. 40:49

    it is the more on equal footing we are again this is why open source effectively wins every time like over period of time open source always wins because it's like it's inspectable everybody can you know trust it they can look at it they can verify that it does what it does and if there's some vulnerability like the the the interesting situation was like some library in that SSH was using right got malicious kind of injection and somebody was testing it in their new upgrade for

94. 41:22

    their database and they found 200 millisecond delay that wasn't there in their Benchmark and like that prevented effectively all of us having zero day vulnerability on SSH where somebody nobody knows who would have Master access to everybody's server right like that's open source how many of the like imagine a Google like somebody working at Google just passed that change and it's like went unnoticed and now

95. 41:46

    somebody else has access to all the Google servers or has access to everything so again this is like security side the you know freedom of speech is the other side right like they're sitting sitting in San Francisco and deciding effectively what in all of the world people can and cannot see and and like it's effectively deciding and and and they kind of required um to do so like they're

96. 42:12

    deciding what effectively all of the world will think right because like they can like so the good example was this is something I've done like I filtered a bunch of data in training the models that were responding on google.com and I did it not because of some like malicious decision but because like the data from Fox News or from Alex Jones website has a lot of conspiracy theory

97. 42:37

    which was hurting my results and evaluation so I just cut out all of that um websites because that you know fixed my model and in turn you know made the model very liberal right not because I'm liberal but because you know I just cut out all of the data that comes from the uh kind of conspiracy websites which somehow mixed with the uh Republican website and so but that's an example of like you know you you effectively um influencing what people going to see and how people going to see it and like some

98. 43:09

    of it can be literally non-malicious just because of some decision- making which you have no idea how that made some of it can be malicious some of this can be you know political view some of this can be us Centric decision to do this or that right which we will not know again Patriot Act is great example of you know government saying there's a uh you know we're going to do it for greater good and then misusing it in all

99. 43:33

    kinds of ways so uh so that's why like again like there is a reason why we're in blockchain space and and actually like Building Technology that's resistant to anyone's influence and visible and everybody can inspect it and even if somebody influenc it you can Fork it and you know get it back on track so similar thing we need for AI similar thing we need to effectively kind of interconnect how we how we process the information

100. 43:57

     how we engaging with uh with this intelligence that's been built um we're now talking about billions of people and trillions of Agents but we also know faster but we also know that getting people and getting agents probably require slightly different things do you see the pth to trillion agents through billion people or they are completely separate things and we can get to trillions of Agents

101. 44:30

     without getting to billion of people so it's a it's a flywheel it's effectively the more users are using the more effectively intents and and requests and asks they have the more asks they have the more agents there needs to be to satisfy this asks and so you kind of have like a and then more things you can do the more people want to use it right it's effectively like I mean we're kind

102. 44:52

     of rebuilding the inventory of Internet through this and the cool thing is the agents themselves they can build themselves too right so and they can make sure they're correct and some agents can be just navigating internet and doing this kind of effectively in the browser some agents you know can replicate Etc so like but we are kind of replicating and re like it's it's the same thing with like uh you know there's trillion web pages on internet so we kind of rebuilding that in a new way

103. 45:23

     yeah can I so yeah so am I right that uh it's a phased Approach at some point it's going to be an agent in itself and agents will propagate agents but in order for us to Kickstart the flywheel we need people that will be deploying these agents and using them and then it's going to have life of its own yeah so I mean we need the assistant inside applications like sweat to actually use this to create requests already that that generates enough usage

104. 45:56

     to to then get a bunch of developers interested in building agents because hey a bunch of sweat users are like asking for this and that and they're willing to pay with sweat here and there so like let me build this let me build that and then now there's more functionality we can integrate into more uh into more applications more traffic more users it also increases sweats retention because people now come in every day because they're like oh it

105. 46:20

     actually can ask you know answer my medical questions that that can you know help me build a diet plan so that I can walk more more and so you can kind of just build that uh yeah fly will thanks um so to have more effective AI agents you need you know data uh from interactions with humans Etc so like how do we as a community look to incentivize the community that we have here now to

106. 46:52

     really align with let's say near AI to be better beta testors um I suppose to to have more effective AI agents what would you say how to to kind of align exactly with with near AI yeah that's actually great question I wouldn't say I have like fully sld out answer yet but that is definitely the direction we need to figure out so again we just launched kind of the chat. neari this is like Alpha version but you can

107. 47:18

     already start you know using it testing it and effectively kind of in this flywheel model like as you as you actually try to try to ask it to do something it may not do it right it has very little functionality right now but that creates enough data to say hey somebody wants this to be done right so let's say like hey I want to tra I want to plan a uh you know travel to Korea for five days

108. 47:44

     and I want you know to do this and that and like right now there's no Integrations with traveling uh kind of uh services and agents and so doesn't do it it's like gives you a plan but it doesn't actually execute and so but that records as like hey there's a travel intent and so now there is interest for developers to build this agent and so we're going to effectively run a hackathon based on the intents people

109. 48:08

     provide to then build those things and then like actually re-engage you know hey now you can actually do it there's like literally a button says you know buy like here's all the tickets you know here's all the hotels and you can do it in one click right so like that's that's a flywheel so if I don't trust open AI with my data why should I trust you or me right because I'm also part of nearer why should they trust why should I trust

110. 48:43

     myself with my data yeah so so the kind of what uh Alex K was talking the new model that's used is called trusted execution environment so this is actually an isolation inside your CPU and effec for GPU as well for like actual compute chip you have that doesn't allow whoever runs the computer to actually inspect what's going on inside and so now you are able to effectively encrypt uh your request send it directly inside that environment so

111. 49:16

     nobody no single person will ever be able to decrypt and access that and then it runs inside an environment the agent runs you know maybe it schedule calls if you actually call falling out into like you know Expedia for example well xedia needs to get your information but from perspective of what happened inside AI inference inside how it process your data Etc all of that stays there and on and kind of encrypted with your key and

112. 49:40

     only you can have have access to it again this is something that we can do it because we have all the kind of blockchain infrastructure to do encryption decryption to have private keys for people and uh we have Micro payments attached to every effectively interaction with this uh secure Enclave and uh so this is all again this is work in progress but this is kind of the model that we see now uh inside this one

113. 50:05

     warning like Intel and Nvidia effectively have you know some ability to uh manufacture like effectively their own chips and so we are trusting them but but like for them to violate how they actually design chips is like a lot bigger problem than you know for example open and I keep confusing you know who's logs are where in their database all right cool uh yeah we have time for one more um hi Leo nice to meet you um we're

114. 50:42

     so we're actually at um tackling the data aggregation and validity problem and we're in your accelerator so thanks for thanks for having us um so part of the decision for us going to near was Frankly Speaking also part of your back background in Ai and compared to I guess all these new AI chains I would love to hear in your own words like what will what do you think will really distinguish near from all the other AI chains that are coming up yeah well thanks for the question the I mean the

115. 51:12

     core is like our commitment to kind of really as I was saying like create this uh uh flywheel of usage and coordinating all of this efforts so the the challenge right now is like everybody's doing a pece meal of products around kind of AI ecosystem right you have data somebody's doing compute somebody's doing training and the thing is like when you go to open AI they do all of this in one right

116. 51:38

     you kind of come in and you have the full product you have the full developer platform you have user platform you have everything and like the challenge with blockchains historic not just even in AI side in blockchain side has been that you know you come and you have like a swap here a lending there a you know like you like I I was trying to find like for UNIS V3 a AO rebalancing tool and I was like where do

117. 52:04

     I find that like it's you cannot find it from Unis swap right you cannot find it like you Google it there's a bunch of stuff like some of the website are closed so what we're trying to do is really put it all together right to create a platform where all of the tools are available all of the data all the compute all of the models inference we creating competitions to attract AI researchers we have you know hackathons for AI agents we're working across you know bit and all us and Quest flow who

118. 52:28

     are actually providing AI Frameworks to really put them together we're putting this into AI assistant integrating it into all the wallets and creating this really kind of interconnected environment where all of this is actually can be successful and so that's kind of the integration the blockchain itself you know can power this for you know billions of transactions it can handle this through intents to like interconnect services so like we're building kind of every single piece or

119. 52:53

     we're working with the ecosystem to build a piece and integrate it into to like cohesive picture amazing thank you all everybody for coming uh that wraps it up for this session really appreciate you all being here I hope you got some value from the time today with Ilia and uh looking forward to seeing you all around the conference thank you thank you for your time