{"id":"7QmkBk3iEpI","title":"Who Decides What Intelligence Becomes?","description":"Who really shapes the future of human–machine systems once capital enters the picture.\n\nNEAR is the unified commerce layer for assets and agents. \n\nWebsite: https://near.org/\nNEAR AI: https://near.ai/\nIntents: https://near.org/intents\nChain Abstraction: https://www.proximity.dev/\nSharding: https://nearone.org/\n\nFor the devs\nGitHub: https://github.com/near \nDev Docs: https://docs.near.org/\nDevhub: https://devhub.near.page/ @NEARDevHub\n\nConnect with socials\nTwitter:   / nearprotocol  \nBlog: https://near.org/blog/\nMedium:   / nearprotocol  \n#NEAR #nearprotocol #NEARAI","channel_name":"NEAR Protocol","published_at":"2026-03-01T00:00:00.000Z","thumbnail_url":"https://img.youtube.com/vi/7QmkBk3iEpI/hqdefault.jpg","transcript_text":"I am excited and delighted to explore this topic with these three very smart individuals on a subject. Uh let's start with some quick intros of what you're focused on and we'll dive deep into like the interplay between capital and AI development. Uh Arnov, kick us off. Cool. Hi everyone. Uh my name is Arnov. I'm a partner at Bankless Ventures. We're an early stage crypto venture capital fund. Hi everyone, my name is Joan Kim. I'm part of firm called fortitude ventures. We're based here in the Bay Area. Um we invest in this thesis called autonomy. Whether you're automating knowledge work, vertical AI or applied AI or physical AI, um embodied AI, which is robotics. Hi everyone, my name is Gor. Uh I'm a new member at the investment team at Hasht uh which is a global early stage venture fund. And I'm Jeff Wilser. I host two podcasts on AI. The people's AI focused on decentralized AI and a general interest AI podcast called AI curious. So as a baselining, let's love your thoughts on kind of largely speaking, how does capital impact the shape of AI development? I know it's an awfully big open-ended question, but how do we get our arms around that? Joanne, what are your thoughts? Um so I think AI has really repriced how software um how how we value software businesses. Um what I mean by this in this in the last 20 years venture revolved around um you know zerocost margins. Um you know you had specific you had margins that were great because you know when you use a product it essentially scaled. Um with AI every time you use the product you incur compute cost. And so this has really flipped how we think about um pure software businesses. And so AI really makes um businesses easier to build but defensibility really hard because now durability is challenged in terms of like a product that maybe took a year to build now it takes weeks and maybe even days. Gor how do you see it? Well, when we started looking into the space, right, you always look at it from a crypto angle first, I guess. And uh when approaching that, the the easiest analysis to do was um how valuable are the moes around the inputs into a new technology, right? And uh when looking at uh you know, frontier models and whatnot, it was relatively easy to deduce when when when all of them kind of took off that um data, compute, etc. are all very high cost items. and and inputs to uh differentiate yourself. And as long as those barriers to entry exist, uh the it's in the incentive set for any developer to continue uh you know leveraging high capital uh to differentiate themselves. And there's no real uh you know other alternative. And so I think that's where we kind of saw the the lack of decentralized solutions take off as compared to centralized ones and why we may have sleepwalked into the same world with AI as we have in the past with internet business models. Um but I'm I'm hopeful and excited to get ahead of things. It's almost like there was so much kind of a head start the centralized players had so much capital already invested there was kind of like not any material real way to be able to catch up at that kind of frontier level. Is that what you're saying? capital was the you know the primary motivator to be differentiated and and what it funded. Arnav yeah I agree with both of you guys. I think the only thing I'll add is just on the relationship between maybe venture capital and AI in that in venture capital really flows where there is very clear speed and scale. So for the first time when you see businesses like Curser or Harvey with growth curves unlike anything we've seen before um and they become unicorns and decacorns and like sub 2year periods uh I think that type of thing will somewhat continue and obviously in many cases we're going to see trade-offs in safety and we'll see tradeoffs in privacy as well. So kind of curious to see how these things play out. Let's have to widen the lens a bit and consider even the big frontier models and the players right as as they are shifting to more focus on like corporate AI I'm wondering how that changes how AI is developed to address the elephant in the room all these I'm going to state the extreme obvious even a lay person perspective here that okay open AI was founded famously as a nonprofit is now very much forprofit Right. It's obviously a very polarizing subject and we could debate that for hours and hours, but I think it's fair to say that good chance that the way they're developing AI might be shifted when there's more capital and economic incentives and pressure for revenue and profit. Uh I view as kind of like a extreme example. How are you guys looking though at how when there is this like focus on having to hit ROI metrics that shifts whether you spend more of your energy on like AGI to benefit humanity versus we need to show third quarter results. I think it's the the easiest analogy to make is early days of Twitter and um how you kind of have an open API that anyone could build on top of the network. the network itself being valuable. In this case, the model itself and and all that it provides is is is where value lies. And as you start uh seeing kind of any application that leverages the you know the infrastructure get valuable, it's in the incentive set for any of these model providers to go and capture that. Um I think we're seeing the same with every valuable application now integrating stables or payments just because it's valuable enough for them to come in. They own distribution and so they're able to close that off. similar to how you know Twitter in the old days kind of shut off the API and uh took everything that that was necessary and so I think naturally no no matter how you frame it you will have these kind of closed off environments develop because it's in their incentive set to do so uh which is why I think something as powerful as you know an AGI powered future or anything related to inherently personal information should exist on chain it's just about the market incentives lining up to make sure that happens you mentioned incentives and that's a great word can you guys talk a more about how incentives shift when there's more capital in play. Um it's fascinating Joanna you said that actually kind of upends the usual incentive to like oh you want you want to show results and build but actually like well the comput is is is costly. Um how else are kind of incentives changing in surprising ways perhaps? Yeah. So we spent a lot of time in vertical or enterprise AI. Um just small anecdote I think uh one big reason with the exception of anthropic or open open AI why we haven't really seen consumer AI applications is um the old ads model like the CPMs just aren't large enough so that they can support a sustainable consumer business today. And so where we found um you know companies with with very stable growth is you know we the biggest question that we've been asking a lot is what is your what is your big mo or like how do you stay defensible and it's one you know you pick really niche or like deep vertical workflows that you can integrate deeply with um so you can capture proprietary data which will help you create a better workflow and like and and and I think um you know openai anthropic they've gone into like different verticals. Um Arnov and I were talking backstage about um Harvey talking about healthcare and so I I think given how fast AI is moving um we're I f focus is all you need right attention is all you need focus all is all you need and I think we're we've seeing that the ROI come from these um measured outcomes versus just you know abstract intelligence or like whatever AGI is and um and when that when that does come. Arnold your thoughts on incentives. Yeah, maybe maybe on a slightly different note than that from a pure venture capital perspective, the thing that's been really interesting is uh not only how obviously capital impacts AI, but how AI impacts capital and specifically what I'm talking about is when we're allocating to new early stage startups and I know every venture capitalist feels this you know what are actually the moes like a technical moat used to be a real thing. your ability to ship code really fast used to be like an actual differentiator. Today now it feels like most things are being commoditized in in many ways. So now we value network effects a lot deeper. We value regulatory modes a lot more. We value uh distribution having a structural edge in customer acquisition. Um so actually the presence of AI makes these things a lot more valuable and I think we index on them a lot heavier. So I think it's an interesting way that actually AI impacts capital allocators. to get back at the spirit of the title of this session to me at core the real fundamental question is actually I think a really powerful one and an important one of okay is there almost a misalignment here or an alignment question or issue of we know that companies need capital and there's like that that is you know no shade it's understandable that is reasonable for a company want capital and reasonable for company to want to um do right by VCs, but I'm guessing the metrics that VCs care about have little to do with like building quote good AI or AI that's fair, AI that's aligned with humanity, right? And I I mean I I'm guessing any spreadsheet you have looking at ROI calculations column F or column G is not have a variable on human alignment. Right? So how are you how do you guys square that? How do you reconcile or how are you thinking about okay all these AI projects need capital they need to show performance but they also hopefully want to not be evil not I'm not at all saying money is evil but you get you see what I'm getting at right there this is kind of like there's there's not really a spreadsheet or metrics on kind of quote good AI how are you thinking about that larger picture yeah I'll go I I think any of the larger AI companies um and their allocators. It's I think they think relatively the same way. The larger AI companies, they're thinking about how do we drive growth? How do we drive scale? How do we drive better gross margins? And in return, they will attract more and more capital. And especially as they size up to these half a trillion dollars, trillion dollar type of data centers, they're going to want to attract these sovereign wealth funds who are going to care more about is this actually a place I can uh you know, put my capital. Um so ultimately I think the only time in which they're really going to slow down and ask like is this for the benefit of humanity will be if the social layer it actually says hey you guys have taken it too far and they start actually canceling their des subscriptions or you know there's a clear impact on the bottom line. So I I fundamentally don't believe that any of the large model providers will self-correct without the social layer demanding it. But we've seen this social I it's a great point. I think in web two we've seen a social layer be very very quick to mobilize right I mean people were look at the sins and original sins of the Facebooks of the world and problems there where social media it's still people are like happy it's like here's my here's my data here's my you know just take it and and we don't need to preserve privacy right so it's like I'm I'm not super optimistic that billions of people will kind of like mobilize in any meaningful way at the social layer But sorry, not a question. That's just my own confession of a of a concern there. Um Joanne, how are you seeing this? How are you see I know it's a complicated uh set of issues here, but how are you seeing that? Yeah, I think on one layer, but from where Arnov was speaking on the application level, I don't think that AI is at or I don't think that humanity is at odds with AI. I actually think it's great. I think it's good for all of us. Um you know, autonomy will create time. Um I I think AI whether it's physical AI or knowledgework AI it's replacing a lot of these you know very expensive or scarce or um you know risky jobs for example in body air robotics um one of our LPs in Korea uh Korea as you know is facing a big age aging um aging population pro uh problem and so for them um finding finding great labor is is a very hard task and so they welcome these solutions because they're having to move their factories closer to the cities and um you know the average age of a lot of these these workers in these factories are in their mid-50s and so um I actually think um there you know it small steps for one big step in into mankind meaning like I don't think this terminator AGI um threat will be in the near term um obviously AI is under overestimate in the overestimated short term underestimated in the long term but I I don't think it's at odds and I think a lot of the applications that we're seeing today are actually very helpful for what um the different problems and tasks that um humans have or humans are facing today and yeah until I'm not a Skynet doomer terminator guy either it's just but there there is I think that the tension between like the the economic incentives for speed versus thoughts of safety and to what extent how that plays out who knows um so in our in our prep uh uh we you mentioned the idea that that speed can monetize faster, right? And there are certain things that are easier to have paybacks on. How has that changed? Even putting aside the Skynet world, right, and back to kind of like planet Earth and reality, how is that shift where we focus our energy on AI when it's true that monetization that's quicker might be more attractive to VCs? Where does that kind of overindex or underindex the space? Well, really quick on on your previous point, um I think the market is is the best determined for for what's the what what's best um and what people want, right? I mean, we can theorize all we want. Um but at the end of the day, what people are paying for or what people use. Um I think DeFi is a great example actually as an analogy. You and and and we're very early for anything internet or AI based too. In finance as an industry has been structurally inefficient for a while. Defi came in as a as an open standard as a competitor that made it more efficient and the market moved to to that solution and and we see it moving onchain which is why you know people claim that blockchain is is inherently a financial solution well it's because it's solving a structurally broken issue. Um and I think at some point the market's going to wake up to the fact that if we've been inherently personal with internet models so far and now you have AI models that are becoming even more personal there's a limit at which we stop. Um and hopefully open standards uh you know kind of take the take the baton from there just takes time. No it's a great point and uh we are legally required to we we have not yet mentioned AI agents and in 15 minutes and that's a a penalty here on me. Um let's talk discoverability. I think uh one made an excellent point earlier uh in our prep that not enough folks are talking about um how agents will be discovered and potentially discovered by other agents and maybe how that could influence um where capital is going and how AI is being built. What are your thoughts on that? Yeah, I can go. Um yeah, I think the agent discoverability thing is super interesting. We were actually talking backstage and you know obviously we had like an app store moment a long time ago and that became the canonical discovery for iOS applications and that built many multiple billions of millions billions of dollar businesses. Um likewise I think on the agent front we're likely to see single interfaces that kind of collapse most of a human's financial interactions and I think how that impacts the entire supply chain is very interesting. Uh, a great example right now, uh, you go on Expedia or Kayak to book your trips, they have a lot of pricing power. They do overcharge you because they know they can because you're so familiar with their interface. They have rewards, there's lock in, all that fun stuff. But if you're actually um, booking a trip through this interface instead, um, that agent is going to go out and find whichever um, whichever intermediary is going to give you the best pricing basically on that trip. And that's great for the end consumer. So I think net net um it's going to lead to better outcomes for the consumer and I do believe that most financial interactions will collapse around one interface and I think if you look at uh somebody like Vlad from Robin Hood he's like very cognizant of um all of the new things that I guess younger folks in particular are looking to do whether it's trade stocks options prediction markets I don't think it's crazy they add an aenic interface and people spend most of their time there sooner than later. So, we went from destroying humanity to better for the consumer. So, it's it's a huge win already. We're already more more optimistic here. I love that. Yeah. I mean, there there are trade-offs, right? Again, like privacy is is not going to be by default. Yeah. Guard rails are still necessary. That's not going to be default, but you know, people are willing to trade speed and scale and for monetization wherever possible. That's a I'm glad you mentioned privacy. I think it's actually a much more to Joan's point, we don't need to be talking about the end of civilization, the scary stuff to to kind of see some real world impacts of possible misalignment. Um, what steps can the space take to ensure that privacy is not something sacrificed or lost uh in the pursuit of capital and how AI is being built to get capital. people have just gotten used to efficiency over privacy if anything and so it's just going to be a market reaction to when people wake up to that fact and I think there's some allocative efficiency to be found between models that are good enough to be run in an open source manner and you have to take the conscious decision to route your requests that way or your preferences that way. Um, and and if if they're more permissionless or or cheaper to do, uh, then it becomes a natural go-to for people. And in all honesty, in the in the in the near future, I don't really see be privacy being a make or break for people who just want to get the job done. Sadly, I completely agree with you. Uh, Joanne, do you want to chime in there? I guess um, one last thought would be I I discoverability really is a distribution problem. Um I think uh you know agents don't have the UI like issue that humans do and so a lot of like brand and UI are qualities that are going to be eroded. Um when I think about how a lot of the problems that AI agents are solving are very obvious and um there's a demand and large need and so it's going to be about really controlling the distribution channels and you know unfortunately I do think a lot of the incumbents are well positioned for that. So if you're a startup builder, um it you're going to have to think about some uni unique ways to sort of um wedge yourself into that into these distribution channels. But I think that's I think I think a lot about distribution when I think about discoverability. You mentioned metrics earlier. would love to hear about I mean I'm actually really fascinated by this how AI feels like a really hard thing to figure out like how and where and when to uh to claim yes like success or not right like it's being a VC being an investor I'm sure it's always complicated to figure out okay early stage where do we define this is working not working and so on like what metrics are most essential to you guys right now and how is has shifted and I'm wondering how it tacks back to like deciding how AI is being built. Arav what for you what's the not to not to show your entire investor playbook but to the extent you're allowed to to chime in at a general level like what types of met how is or maybe I'll ask it differently how are AI metrics different from more kind of traditional conventional metrics yeah I I understand um obviously on the AI front we're seeing a lot of explosive growth super quickly I think quantifying the quality of that revenue is actually super difficult right And also again the inference costs are super high so CPM matters a lot as well. Um so again I I don't invest purely in AI stuff so I'm not best qualified to answer that. Um but yeah it's it's completely different and I know for a fact a lot of my friends who you know are investing in pure AI. uh evaluating the quality and the durability and the scalability of that revenue is increasingly difficult especially when a lot of things have pretty serious growth curves coming out the gate and it's very difficult to determine whether that's going to expand or not and Joann's point the kind of paradox is it could be they're they're getting real traction but in cost are high if they're bleeding money you're like oh is this good or is this bad it must be a little complicated as an investor to to judge that. Yeah. Yeah. No, it's it's super hard for sure. Yeah. On the flip side, I I wasn't sure on the question, but like the only metric that VCs care about is DPI. How much money did you return to your investors? Yeah, Gor, how about yourself? Uh I think there's a kind of a notional value of creativity attached to each of these startups because um again being in the crypto space you kind of make analogies back and I think there's certain conflationary trends going on with what people are building in the AI space right now and how liquidity used to hop around in the early days of DeFi when you had one protocol and you have you know the the code open source and so a competitor would build a slightly better version. you kind of have a similar dynamic of AI uh tools being used and then of course the looming threat that open AAI or any model provider kind of either turns off the pivot or just vertically integrates that use case. So I think uh there's there's this notional value of creativity which makes uh essentially the question is how hard is your company replicable? Yeah. How how hard is it to replicate? And back to moat right is almost a creative mode in a sense. Essentially, it's like, you know, are you applying the the models in a way which is creative enough such that it can stand on its own and there isn't like a fundamental dependency on the base layer. Much easier much harder said than done. Yeah. Uh, as we wrap up, would love some predictions from you all on how this shakes out and you could take that however you'd like, but broadly speaking on the intersection of capital and AI and how AI is being developed. uh time frame of your choice. Could be a month from now, could be 20 years from now. Dealer's choice. Uh Arnov, any predictions for the future here? Yeah, I I don't know how many people read there was a very recent article that Catrini wrote on Twitter, like 50 million views this past weekend actually. Um and it was like looking back, we're in 2028, 2030 looking back, how have things changed? Um and I think maybe over a longer term horizon, not even that long, like 5 to 8 years. Um I think this is the first time where technology will displace jobs but not create an equivalent amount. And I think the way that impacts the economy will be super interesting because it also well increases the gross margins for corporations. It does decrease the amount of marginal buyers and that obviously creates a new paradigm in which we have to uh you know settle for. So I guess that's my prediction. Interesting is a very euphemistic word right there. Uh Joanne, um I agree with both of what they said. Um I think more than uh like knowledge work AI, I'm really interested to see how embodied AI or robotics sort of permeates our lives. Um uh at Fortitude, we were thinking a lot about like what does the future look like? Is it humanoids where you know your humanoid robot that was doing dishes and cooking you food, stepping out to drive your car? Um probably not. It's probably going to be a lot of like these specialized robots where a Whimo or or a Tesla drives you in the car and then you have specific robots for um you know making food or like cleaning the house. And so um I I'm sort of excited for what is going to be happening after the next unlock where when the hardware starts to enter the picture. Um and so you know excited to speak with anybody in the audience also who has ideas in in this physical embodied space as well. And I agree with both your and agree with your prediction as well. For better or worse, I I agree with you. Uh final thoughts on predictions. Um in the short term, I think uh what what DeFi did for financial expressivity, AI tools are doing for creativity or expressivity with building. And so while building in the crypto blockchain space seems uh very negative EV right now, I think we're very optimistic actually because now anyone has the power and the tools to continue building um and and and the barriers of entry have just gotten lower. Um the the area in which they express that in is is open to them, but it uh you know crypto and blockchain can be a very valuable place to express that interest. Um down the line I um I think my my very long-term view is that we live in a world where you know privacy is inherent and you have user own data which is which is transferable and now memory and intelligence become even more inherently personal. So that is uh you know also transferable between services and we hold control of it. Do I think that's happening anytime soon? Um probably not. I think we're sleepwalking into yet another iteration of internet owned business model. Love your long-term vision. A prediction of mine here. I think we'll see more capital shifting into robots, but not necessarily humanoid robots, more specialized robots that will become more and more mainstream in the future. Uh, on that note, we're out of time. Please join me in thanking our panelists. Thank you guys.","transcript_synced_at":"2026-11-03T04:38:06.151Z","segments":[{"start_seconds":9,"text":"I am excited and delighted to explore this topic with these three very smart individuals on a subject. Uh let's start","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=9s"},{"start_seconds":17,"text":"with some quick intros of what you're focused on and we'll dive deep into like the interplay between capital and AI development. Uh Arnov, kick us off.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=17s"},{"start_seconds":25,"text":"Cool. Hi everyone. Uh my name is Arnov. I'm a partner at Bankless Ventures.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=25s"},{"start_seconds":29,"text":"We're an early stage crypto venture capital fund.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=29s"},{"start_seconds":33,"text":"Hi everyone, my name is Joan Kim. I'm part of firm called fortitude ventures.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=33s"},{"start_seconds":37,"text":"We're based here in the Bay Area. Um we invest in this thesis called autonomy.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=37s"},{"start_seconds":41,"text":"Whether you're automating knowledge work, vertical AI or applied AI or physical AI, um embodied AI, which is robotics.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=41s"},{"start_seconds":49,"text":"Hi everyone, my name is Gor. Uh I'm a new member at the investment team at Hasht uh which is a global early stage venture fund.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=49s"},{"start_seconds":56,"text":"And I'm Jeff Wilser. I host two podcasts on AI. The people's AI focused on decentralized AI and a general interest","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=56s"},{"start_seconds":64,"text":"AI podcast called AI curious. So as a baselining, let's love your thoughts on","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=64s"},{"start_seconds":71,"text":"kind of largely speaking, how does capital impact the shape of AI development? I know it's an awfully big open-ended question, but how do we get","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=71s"},{"start_seconds":80,"text":"our arms around that? Joanne, what are your thoughts?","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=80s"},{"start_seconds":82,"text":"Um so I think AI has really repriced how software um how how we value software businesses. Um what I mean by this in","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=82s"},{"start_seconds":91,"text":"this in the last 20 years venture revolved around um you know zerocost margins. Um you know you had specific","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=91s"},{"start_seconds":100,"text":"you had margins that were great because you know when you use a product it essentially scaled. Um with AI every","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=100s"},{"start_seconds":107,"text":"time you use the product you incur compute cost. And so this has really flipped how we think about um pure software businesses. And so AI really","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=107s"},{"start_seconds":116,"text":"makes um businesses easier to build but defensibility really hard because now durability is challenged in terms of","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=116s"},{"start_seconds":123,"text":"like a product that maybe took a year to build now it takes weeks and maybe even days.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=123s"},{"start_seconds":128,"text":"Gor how do you see it? Well, when we started looking into the space, right, you always look at it from a crypto angle first, I guess. And uh when","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=128s"},{"start_seconds":136,"text":"approaching that, the the easiest analysis to do was um how valuable are the moes around the inputs into a new","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=136s"},{"start_seconds":145,"text":"technology, right? And uh when looking at uh you know, frontier models and whatnot, it was relatively easy to deduce when when when all of them kind of took off that um data, compute, etc.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=145s"},{"start_seconds":156,"text":"are all very high cost items. and and inputs to uh differentiate yourself. And","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=156s"},{"start_seconds":164,"text":"as long as those barriers to entry exist, uh the it's in the incentive set for any developer to continue uh you","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=164s"},{"start_seconds":171,"text":"know leveraging high capital uh to differentiate themselves. And there's no real uh you know other alternative. And so I think that's where we kind of saw","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=171s"},{"start_seconds":179,"text":"the the lack of decentralized solutions take off as compared to centralized ones and why we may have sleepwalked into the same world with AI as we have in the","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=179s"},{"start_seconds":188,"text":"past with internet business models. Um but I'm I'm hopeful and excited to get ahead of things. It's almost like there was so much kind of a head start","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=188s"},{"start_seconds":197,"text":"the centralized players had so much capital already invested there was kind of like not any material real way to be able to catch up at that kind of","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=197s"},{"start_seconds":204,"text":"frontier level. Is that what you're saying? capital was the you know the primary motivator to be differentiated and and what it funded.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=204s"},{"start_seconds":211,"text":"Arnav yeah I agree with both of you guys. I think the only thing I'll add is just on the relationship between maybe venture","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=211s"},{"start_seconds":218,"text":"capital and AI in that in venture capital really flows where there is very clear speed and scale. So for the first","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=218s"},{"start_seconds":226,"text":"time when you see businesses like Curser or Harvey with growth curves unlike anything we've seen before um and they become unicorns and decacorns and like","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=226s"},{"start_seconds":234,"text":"sub 2year periods uh I think that type of thing will somewhat continue and obviously in many cases we're going to","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=234s"},{"start_seconds":242,"text":"see trade-offs in safety and we'll see tradeoffs in privacy as well. So kind of curious to see how these things play","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=242s"},{"start_seconds":249,"text":"out. Let's have to widen the lens a bit and consider even the big frontier models and the players right as","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=249s"},{"start_seconds":257,"text":"as they are shifting to more focus on like corporate AI I'm wondering how that changes how AI is developed to address","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=257s"},{"start_seconds":265,"text":"the elephant in the room all these I'm going to state the extreme obvious even a lay person perspective here that okay","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=265s"},{"start_seconds":272,"text":"open AI was founded famously as a nonprofit is now very much forprofit","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=272s"},{"start_seconds":280,"text":"Right. It's obviously a very polarizing subject and we could debate that for hours and hours, but I think it's fair to say that good chance that the way","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=280s"},{"start_seconds":288,"text":"they're developing AI might be shifted when there's more capital and economic incentives and pressure for revenue and","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=288s"},{"start_seconds":296,"text":"profit. Uh I view as kind of like a extreme example. How are you guys looking though at how when there is this like focus on having to hit ROI metrics","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=296s"},{"start_seconds":306,"text":"that shifts whether you spend more of your energy on like AGI to benefit humanity versus we need to show third quarter results.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=306s"},{"start_seconds":316,"text":"I think it's the the easiest analogy to make is early days of Twitter and um how you kind of have an open API that anyone","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=316s"},{"start_seconds":324,"text":"could build on top of the network. the network itself being valuable. In this case, the model itself and and all that it provides is is is where value lies.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=324s"},{"start_seconds":332,"text":"And as you start uh seeing kind of any application that leverages the you know the infrastructure get valuable, it's in the incentive set for any of these model","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=332s"},{"start_seconds":341,"text":"providers to go and capture that. Um I think we're seeing the same with every valuable application now integrating stables or payments just because it's valuable enough for them to come in.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=341s"},{"start_seconds":351,"text":"They own distribution and so they're able to close that off. similar to how you know Twitter in the old days kind of shut off the API and uh took everything","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=351s"},{"start_seconds":358,"text":"that that was necessary and so I think naturally no no matter how you frame it you will have these kind of closed off environments develop because it's in","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=358s"},{"start_seconds":366,"text":"their incentive set to do so uh which is why I think something as powerful as you know an AGI powered future or anything","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=366s"},{"start_seconds":373,"text":"related to inherently personal information should exist on chain it's just about the market incentives lining up to make sure that happens you mentioned incentives and that's a","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=373s"},{"start_seconds":382,"text":"great word can you guys talk a more about how incentives shift when there's more capital in play. Um it's fascinating Joanna you said that","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=382s"},{"start_seconds":390,"text":"actually kind of upends the usual incentive to like oh you want you want to show results and build but actually","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=390s"},{"start_seconds":396,"text":"like well the comput is is is costly. Um how else are kind of incentives changing in surprising ways perhaps?","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=396s"},{"start_seconds":403,"text":"Yeah. So we spent a lot of time in vertical or enterprise AI. Um just small anecdote I think uh one big reason with","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=403s"},{"start_seconds":412,"text":"the exception of anthropic or open open AI why we haven't really seen consumer AI applications is um the old ads model","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=412s"},{"start_seconds":419,"text":"like the CPMs just aren't large enough so that they can support a sustainable consumer business today. And so where we","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=419s"},{"start_seconds":426,"text":"found um you know companies with with very stable growth is you know we the biggest question that we've been asking a lot is what is your what is your big","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=426s"},{"start_seconds":435,"text":"mo or like how do you stay defensible and it's one you know you pick really niche or like deep vertical workflows that you can integrate deeply with um so","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=435s"},{"start_seconds":444,"text":"you can capture proprietary data which will help you create a better workflow and like and and and I think um you know","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=444s"},{"start_seconds":452,"text":"openai anthropic they've gone into like different verticals. Um Arnov and I were talking backstage about um Harvey talking about healthcare and so I I","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=452s"},{"start_seconds":461,"text":"think given how fast AI is moving um we're I f focus is all you need right attention is all you need focus all is","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=461s"},{"start_seconds":469,"text":"all you need and I think we're we've seeing that the ROI come from these um measured outcomes versus just you know","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=469s"},{"start_seconds":476,"text":"abstract intelligence or like whatever AGI is and um and when that when that does come. Arnold your thoughts on incentives.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=476s"},{"start_seconds":484,"text":"Yeah, maybe maybe on a slightly different note than that from a pure venture capital perspective, the thing that's been really interesting is uh not","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=484s"},{"start_seconds":492,"text":"only how obviously capital impacts AI, but how AI impacts capital and specifically what I'm talking about is when we're allocating to new early stage","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=492s"},{"start_seconds":501,"text":"startups and I know every venture capitalist feels this you know what are actually the moes like a technical moat used to be a real thing. your ability to","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=501s"},{"start_seconds":508,"text":"ship code really fast used to be like an actual differentiator. Today now it feels like most things are being commoditized in in many ways. So now we","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=508s"},{"start_seconds":517,"text":"value network effects a lot deeper. We value regulatory modes a lot more. We value uh distribution having a structural edge in customer acquisition.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=517s"},{"start_seconds":526,"text":"Um so actually the presence of AI makes these things a lot more valuable and I think we index on them a lot heavier. So I think it's an interesting way that actually AI impacts capital allocators.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=526s"},{"start_seconds":537,"text":"to get back at the spirit of the title of this session to me at core the real fundamental question is actually I think","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=537s"},{"start_seconds":546,"text":"a really powerful one and an important one of okay is there almost a misalignment here or an alignment","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=546s"},{"start_seconds":553,"text":"question or issue of we know that companies need capital and there's like that that is you know no shade it's","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=553s"},{"start_seconds":562,"text":"understandable that is reasonable for a company want capital and reasonable for company to want to um do right by VCs,","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=562s"},{"start_seconds":570,"text":"but I'm guessing the metrics that VCs care about have little to do with like","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=570s"},{"start_seconds":577,"text":"building quote good AI or AI that's fair, AI that's aligned with humanity,","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=577s"},{"start_seconds":584,"text":"right? And I I mean I I'm guessing any spreadsheet you have looking at ROI","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=584s"},{"start_seconds":591,"text":"calculations column F or column G is not have a variable on human alignment. Right? So how are you how do you guys square that?","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=591s"},{"start_seconds":602,"text":"How do you reconcile or how are you thinking about okay all these AI projects need capital they need to show","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=602s"},{"start_seconds":610,"text":"performance but they also hopefully want to not be evil not I'm not at all saying money is evil but you get you see what I'm","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=610s"},{"start_seconds":617,"text":"getting at right there this is kind of like there's there's not really a spreadsheet or metrics on kind of quote","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=617s"},{"start_seconds":624,"text":"good AI how are you thinking about that larger picture yeah I'll go I I think any of the larger AI companies","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=624s"},{"start_seconds":632,"text":"um and their allocators. It's I think they think relatively the same way. The larger AI companies, they're thinking about how do we drive growth? How do we","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=632s"},{"start_seconds":639,"text":"drive scale? How do we drive better gross margins? And in return, they will attract more and more capital. And especially as they size up to these half","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=639s"},{"start_seconds":646,"text":"a trillion dollars, trillion dollar type of data centers, they're going to want to attract these sovereign wealth funds who are going to care more about is this","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=646s"},{"start_seconds":654,"text":"actually a place I can uh you know, put my capital. Um so ultimately I think the only time in which they're really going to slow down and ask like is this for","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=654s"},{"start_seconds":662,"text":"the benefit of humanity will be if the social layer it actually says hey you guys have taken it too far and they start actually canceling their des","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=662s"},{"start_seconds":671,"text":"subscriptions or you know there's a clear impact on the bottom line. So I I fundamentally don't believe that any of","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=671s"},{"start_seconds":678,"text":"the large model providers will self-correct without the social layer demanding it. But we've seen this social","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=678s"},{"start_seconds":685,"text":"I it's a great point. I think in web two we've seen a social layer be very very quick to mobilize right I mean people were look at the sins and original sins","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=685s"},{"start_seconds":694,"text":"of the Facebooks of the world and problems there where social media it's still people are like happy it's like","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=694s"},{"start_seconds":702,"text":"here's my here's my data here's my you know just take it and and we don't need to preserve privacy right so it's like I'm I'm not super optimistic that","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=702s"},{"start_seconds":711,"text":"billions of people will kind of like mobilize in any meaningful way at the social layer But sorry, not a question.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=711s"},{"start_seconds":718,"text":"That's just my own confession of a of a concern there. Um Joanne, how are you seeing this? How are you see I know it's a complicated uh","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=718s"},{"start_seconds":726,"text":"set of issues here, but how are you seeing that?","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=726s"},{"start_seconds":728,"text":"Yeah, I think on one layer, but from where Arnov was speaking on the application level, I don't think that AI","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=728s"},{"start_seconds":735,"text":"is at or I don't think that humanity is at odds with AI. I actually think it's great. I think it's good for all of us.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=735s"},{"start_seconds":742,"text":"Um you know, autonomy will create time.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=742s"},{"start_seconds":745,"text":"Um I I think AI whether it's physical AI or knowledgework AI it's replacing a lot of these you know very expensive or","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=745s"},{"start_seconds":754,"text":"scarce or um you know risky jobs for example in body air robotics um one of our LPs in Korea uh Korea as you know is","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=754s"},{"start_seconds":762,"text":"facing a big age aging um aging population pro uh problem and so for them um finding finding great labor is","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=762s"},{"start_seconds":770,"text":"is a very hard task and so they welcome these solutions because they're having to move their factories closer to the cities and um you know the average age","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=770s"},{"start_seconds":778,"text":"of a lot of these these workers in these factories are in their mid-50s and so um I actually think um there you know it","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=778s"},{"start_seconds":788,"text":"small steps for one big step in into mankind meaning like I don't think this terminator AGI um threat will be in the","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=788s"},{"start_seconds":797,"text":"near term um obviously AI is under overestimate in the overestimated short term underestimated in the long term but","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=797s"},{"start_seconds":804,"text":"I I don't think it's at odds and I think a lot of the applications that we're seeing today are actually very helpful for what um the different problems and","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=804s"},{"start_seconds":811,"text":"tasks that um humans have or humans are facing today and yeah until I'm not a Skynet doomer terminator guy either it's just but","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=811s"},{"start_seconds":819,"text":"there there is I think that the tension between like the the economic incentives for speed versus thoughts of safety and","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=819s"},{"start_seconds":827,"text":"to what extent how that plays out who knows um so in our in our prep uh uh we you mentioned the idea that that speed","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=827s"},{"start_seconds":835,"text":"can monetize faster, right? And there are certain things that are easier to have paybacks on. How has that changed?","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=835s"},{"start_seconds":841,"text":"Even putting aside the Skynet world, right, and back to kind of like planet Earth and reality, how is that shift where we focus our energy on AI when","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=841s"},{"start_seconds":850,"text":"it's true that monetization that's quicker might be more attractive to VCs?","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=850s"},{"start_seconds":857,"text":"Where does that kind of overindex or underindex the space?","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=857s"},{"start_seconds":861,"text":"Well, really quick on on your previous point, um I think the market is is the best determined for for what's the what","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=861s"},{"start_seconds":869,"text":"what's best um and what people want, right? I mean, we can theorize all we want. Um but at the end of the day, what people are paying for or what people use.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=869s"},{"start_seconds":877,"text":"Um I think DeFi is a great example actually as an analogy. You and and and we're very early for anything internet or AI based too. In finance as an","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=877s"},{"start_seconds":886,"text":"industry has been structurally inefficient for a while. Defi came in as a as an open standard as a competitor that made it more efficient and the","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=886s"},{"start_seconds":894,"text":"market moved to to that solution and and we see it moving onchain which is why you know people claim that blockchain is is inherently a financial solution well","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=894s"},{"start_seconds":904,"text":"it's because it's solving a structurally broken issue. Um and I think at some point the market's going to wake up to the fact that if we've been inherently","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=904s"},{"start_seconds":911,"text":"personal with internet models so far and now you have AI models that are becoming even more personal there's a limit at which we stop. Um and hopefully open","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=911s"},{"start_seconds":919,"text":"standards uh you know kind of take the take the baton from there just takes time.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=919s"},{"start_seconds":924,"text":"No it's a great point and uh we are legally required to we we have not yet mentioned AI agents and in 15 minutes","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=924s"},{"start_seconds":931,"text":"and that's a a penalty here on me. Um let's talk discoverability. I think uh one made an excellent point earlier uh","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=931s"},{"start_seconds":940,"text":"in our prep that not enough folks are talking about um how agents will be discovered and potentially discovered by","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=940s"},{"start_seconds":948,"text":"other agents and maybe how that could influence um where capital is going and how AI is being built. What are your thoughts on that?","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=948s"},{"start_seconds":957,"text":"Yeah, I can go. Um yeah, I think the agent discoverability thing is super interesting. We were actually talking backstage and you know obviously we had","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=957s"},{"start_seconds":965,"text":"like an app store moment a long time ago and that became the canonical discovery for iOS applications and that built many multiple billions of millions billions","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=965s"},{"start_seconds":974,"text":"of dollar businesses. Um likewise I think on the agent front we're likely to see single interfaces that kind of collapse most of a human's financial","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=974s"},{"start_seconds":983,"text":"interactions and I think how that impacts the entire supply chain is very interesting. Uh, a great example right now, uh, you go on Expedia or Kayak to","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=983s"},{"start_seconds":992,"text":"book your trips, they have a lot of pricing power. They do overcharge you because they know they can because you're so familiar with their interface.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=992s"},{"start_seconds":997,"text":"They have rewards, there's lock in, all that fun stuff. But if you're actually um, booking a trip through this interface instead, um, that agent is","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=997s"},{"start_seconds":1005,"text":"going to go out and find whichever um, whichever intermediary is going to give you the best pricing basically on that trip. And that's great for the end","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1005s"},{"start_seconds":1014,"text":"consumer. So I think net net um it's going to lead to better outcomes for the consumer and I do believe that most financial interactions will collapse","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1014s"},{"start_seconds":1021,"text":"around one interface and I think if you look at uh somebody like Vlad from Robin Hood he's like very cognizant of um all","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1021s"},{"start_seconds":1029,"text":"of the new things that I guess younger folks in particular are looking to do whether it's trade stocks options prediction markets I don't think it's","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1029s"},{"start_seconds":1036,"text":"crazy they add an aenic interface and people spend most of their time there sooner than later. So, we went from destroying humanity to better for the consumer. So, it's it's a huge win","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1036s"},{"start_seconds":1045,"text":"already. We're already more more optimistic here. I love that.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1045s"},{"start_seconds":1047,"text":"Yeah. I mean, there there are trade-offs, right? Again, like privacy is is not going to be by default. Yeah.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1047s"},{"start_seconds":1053,"text":"Guard rails are still necessary. That's not going to be default, but you know, people are willing to trade speed and scale and for monetization wherever possible.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1053s"},{"start_seconds":1061,"text":"That's a I'm glad you mentioned privacy.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1061s"},{"start_seconds":1062,"text":"I think it's actually a much more to Joan's point, we don't need to be talking about the end of civilization, the scary stuff to to kind of see some","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1062s"},{"start_seconds":1070,"text":"real world impacts of possible misalignment. Um, what steps can the space take to ensure that privacy is not","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1070s"},{"start_seconds":1078,"text":"something sacrificed or lost uh in the pursuit of capital and how AI is being built to get capital.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1078s"},{"start_seconds":1085,"text":"people have just gotten used to efficiency over privacy if anything and so it's just going to be a market reaction to when people wake up to that","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1085s"},{"start_seconds":1094,"text":"fact and I think there's some allocative efficiency to be found between models that are good enough to be run in an open source manner and you have to take","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1094s"},{"start_seconds":1102,"text":"the conscious decision to route your requests that way or your preferences that way. Um, and and if if they're more","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1102s"},{"start_seconds":1110,"text":"permissionless or or cheaper to do, uh, then it becomes a natural go-to for people. And in all honesty, in the in the in the near future, I don't really","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1110s"},{"start_seconds":1118,"text":"see be privacy being a make or break for people who just want to get the job done.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1118s"},{"start_seconds":1123,"text":"Sadly, I completely agree with you. Uh, Joanne, do you want to chime in there?","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1123s"},{"start_seconds":1127,"text":"I guess um, one last thought would be I I discoverability really is a distribution problem. Um I think uh you","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1127s"},{"start_seconds":1137,"text":"know agents don't have the UI like issue that humans do and so a lot of like brand and UI are qualities that are going to be eroded. Um when I think","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1137s"},{"start_seconds":1146,"text":"about how a lot of the problems that AI agents are solving are very obvious and um there's a demand and large need and","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1146s"},{"start_seconds":1154,"text":"so it's going to be about really controlling the distribution channels and you know unfortunately I do think a lot of the incumbents are well positioned for that. So if you're a","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1154s"},{"start_seconds":1162,"text":"startup builder, um it you're going to have to think about some uni unique ways to sort of um wedge yourself into that","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1162s"},{"start_seconds":1170,"text":"into these distribution channels. But I think that's I think I think a lot about distribution when I think about discoverability.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1170s"},{"start_seconds":1175,"text":"You mentioned metrics earlier. would love to hear about I mean I'm actually really fascinated by this how AI feels","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1175s"},{"start_seconds":1183,"text":"like a really hard thing to figure out like how and where and when to uh to","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1183s"},{"start_seconds":1192,"text":"claim yes like success or not right like it's being a VC being an investor I'm sure it's always complicated to figure","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1192s"},{"start_seconds":1199,"text":"out okay early stage where do we define this is working not working and so on like what metrics are most essential to you guys right now and how is has","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1199s"},{"start_seconds":1207,"text":"shifted and I'm wondering how it tacks back to like deciding how AI is being built.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1207s"},{"start_seconds":1213,"text":"Arav what for you what's the not to not to show your entire investor playbook but to the extent you're allowed to to chime in at a general level like what","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1213s"},{"start_seconds":1221,"text":"types of met how is or maybe I'll ask it differently how are AI metrics different from more kind of traditional conventional metrics","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1221s"},{"start_seconds":1229,"text":"yeah I I understand um obviously on the AI front we're seeing a lot of explosive growth super quickly I think quantifying","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1229s"},{"start_seconds":1238,"text":"the quality of that revenue is actually super difficult right And also again the inference costs are super high so CPM","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1238s"},{"start_seconds":1245,"text":"matters a lot as well. Um so again I I don't invest purely in AI stuff so I'm not best qualified to answer that. Um","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1245s"},{"start_seconds":1254,"text":"but yeah it's it's completely different and I know for a fact a lot of my friends who you know are investing in pure AI. uh evaluating the quality and","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1254s"},{"start_seconds":1261,"text":"the durability and the scalability of that revenue is increasingly difficult especially when a lot of things have pretty serious growth curves coming out","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1261s"},{"start_seconds":1268,"text":"the gate and it's very difficult to determine whether that's going to expand or not and Joann's point the kind of paradox is it could be they're they're","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1268s"},{"start_seconds":1276,"text":"getting real traction but in cost are high if they're bleeding money you're like oh is this good or is this bad it must be a little complicated as an investor to to judge that.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1276s"},{"start_seconds":1285,"text":"Yeah. Yeah. No, it's it's super hard for sure.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1285s"},{"start_seconds":1288,"text":"Yeah. On the flip side, I I wasn't sure on the question, but like the only metric that VCs care about is DPI. How much money did you return to your investors?","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1288s"},{"start_seconds":1297,"text":"Yeah, Gor, how about yourself? Uh I think there's a kind of a notional value of creativity attached to each of these","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1297s"},{"start_seconds":1303,"text":"startups because um again being in the crypto space you kind of make analogies back and I think there's certain conflationary trends going on with what","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1303s"},{"start_seconds":1312,"text":"people are building in the AI space right now and how liquidity used to hop around in the early days of DeFi when you had one protocol and you have you","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1312s"},{"start_seconds":1320,"text":"know the the code open source and so a competitor would build a slightly better version. you kind of have a similar dynamic of AI uh tools being used and","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1320s"},{"start_seconds":1329,"text":"then of course the looming threat that open AAI or any model provider kind of either turns off the pivot or just vertically integrates that use case. So","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1329s"},{"start_seconds":1337,"text":"I think uh there's there's this notional value of creativity which makes uh essentially the question is how hard is your company replicable?","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1337s"},{"start_seconds":1346,"text":"Yeah. How how hard is it to replicate?","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1346s"},{"start_seconds":1348,"text":"And back to moat right is almost a creative mode in a sense. Essentially, it's like, you know, are you applying the the models in a way which is creative enough","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1348s"},{"start_seconds":1357,"text":"such that it can stand on its own and there isn't like a fundamental dependency on the base layer. Much easier much harder said than done.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1357s"},{"start_seconds":1364,"text":"Yeah. Uh, as we wrap up, would love some predictions from you all on how this shakes out and you could take that however you'd like, but broadly speaking","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1364s"},{"start_seconds":1372,"text":"on the intersection of capital and AI and how AI is being developed. uh time frame of your choice. Could be a month from now, could be 20 years from now.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1372s"},{"start_seconds":1382,"text":"Dealer's choice. Uh Arnov, any predictions for the future here?","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1382s"},{"start_seconds":1386,"text":"Yeah, I I don't know how many people read there was a very recent article that Catrini wrote on Twitter, like 50 million views this past weekend","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1386s"},{"start_seconds":1393,"text":"actually. Um and it was like looking back, we're in 2028, 2030 looking back, how have things changed? Um and I think","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1393s"},{"start_seconds":1401,"text":"maybe over a longer term horizon, not even that long, like 5 to 8 years. Um I think this is the first time where","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1401s"},{"start_seconds":1408,"text":"technology will displace jobs but not create an equivalent amount. And I think the way that impacts the economy will be","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1408s"},{"start_seconds":1416,"text":"super interesting because it also well increases the gross margins for corporations. It does decrease the amount of marginal buyers and that","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1416s"},{"start_seconds":1424,"text":"obviously creates a new paradigm in which we have to uh you know settle for. So I guess that's my prediction.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1424s"},{"start_seconds":1431,"text":"Interesting is a very euphemistic word right there. Uh Joanne, um I agree with both of what they said.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1431s"},{"start_seconds":1438,"text":"Um I think more than uh like knowledge work AI, I'm really interested to see how embodied AI or robotics sort of","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1438s"},{"start_seconds":1446,"text":"permeates our lives. Um uh at Fortitude, we were thinking a lot about like what does the future look like? Is it humanoids where you know your humanoid","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1446s"},{"start_seconds":1453,"text":"robot that was doing dishes and cooking you food, stepping out to drive your car? Um probably not. It's probably going to be a lot of like these","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1453s"},{"start_seconds":1461,"text":"specialized robots where a Whimo or or a Tesla drives you in the car and then you have specific robots for um you know making food or like cleaning the house.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1461s"},{"start_seconds":1470,"text":"And so um I I'm sort of excited for what is going to be happening after the next unlock where when the hardware starts to","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1470s"},{"start_seconds":1478,"text":"enter the picture. Um and so you know excited to speak with anybody in the audience also who has ideas in in this","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1478s"},{"start_seconds":1486,"text":"physical embodied space as well. And I agree with both your and agree with your prediction as well. For better or worse, I I agree with you. Uh final thoughts on predictions.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1486s"},{"start_seconds":1495,"text":"Um in the short term, I think uh what what DeFi did for financial expressivity, AI tools are doing for","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1495s"},{"start_seconds":1503,"text":"creativity or expressivity with building. And so while building in the crypto blockchain space seems uh very negative EV right now, I think we're","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1503s"},{"start_seconds":1512,"text":"very optimistic actually because now anyone has the power and the tools to continue building um and and and the barriers of entry have just gotten","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1512s"},{"start_seconds":1520,"text":"lower. Um the the area in which they express that in is is open to them, but it uh you know crypto and blockchain can be a very valuable place to express that","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1520s"},{"start_seconds":1529,"text":"interest. Um down the line I um I think my my very long-term view is that we live in a world where you know privacy","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1529s"},{"start_seconds":1537,"text":"is inherent and you have user own data which is which is transferable and now memory and intelligence become even more inherently personal. So that is uh you","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1537s"},{"start_seconds":1545,"text":"know also transferable between services and we hold control of it. Do I think that's happening anytime soon? Um probably not. I think we're sleepwalking","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1545s"},{"start_seconds":1553,"text":"into yet another iteration of internet owned business model.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1553s"},{"start_seconds":1556,"text":"Love your long-term vision. A prediction of mine here. I think we'll see more capital shifting into robots, but not necessarily humanoid robots, more","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1556s"},{"start_seconds":1563,"text":"specialized robots that will become more and more mainstream in the future. Uh, on that note, we're out of time. Please join me in thanking our panelists. Thank you guys.","url":"https://www.youtube.com/watch?v=7QmkBk3iEpI&t=1563s"}],"youtube_url":"https://youtube.com/watch?v=7QmkBk3iEpI","url":"https://near-pedia.xyz/videos/7QmkBk3iEpI"}