Sam Altman’s Answer to Bots

[00:00] - I think we have some big news today. Can you please address the rumors? - Which one? [00:04] You don't know which one? No, I don't know which one. What happened? What happened? Well, I think we're announcing $250 million. With investment from OpenAI, how are you going to be working with them? It's a direct investment from OpenAI, not the venture arm. And so it is an actual collaboration. Three major areas, I think, will be biology, chips, AI systems. Brain control. [00:27] Not brain control. Like world, you like to go after controversial spaces. Biometrics has also perked some ears up. What are the biggest misconceptions? The biggest one is, of course, around privacy. What's so counterintuitive about what we're doing here is that we have designed a system that can be fundamentally anonymous and is fundamentally anonymous and privacy-deserving. And biometrics, I think, is the only way to actually solve the problem. Platforms like X will not be usable if they don't come up with completely new technical solutions. [00:57] to combat the bots. [00:58] It's not about combating the bots, because that will be a lost cause, but rather every human on these platforms will need some version of a proof of human and actually prove that their content is authentic. And then we will have all the rest. [01:11] The goal of this is to create the largest real human network on the internet because we think that will actually really matter because I think 99.9% or something of all internet traffic will be AI driven.

[01:24] Alex Polania, welcome to Sorcery. Thank you for having me. So for people who don't know what world is, what is world and what is the full stack that you're building? Yeah, so it is the real human network. That's how we were talking about it. [01:40] So we're trying to imagine what real humans on the internet will need. And so it starts with a proof of human. And so that's, these are actually the devices that do that. But the simple, simple problem statement is that, as I mentioned before, AI has the Turing test and it can can look like a human, it can talk like human on the internet. And so we will need a proof of human for every place that we interact with other people. And that turns out to be a quite [02:10] challenging problem. Like it needs to be privacy preserving, it needs to be able to work in every country. So there's like, there's like many things around that that we had to solve. And, you know, I think we came up with probably current technology, one of the best, the best ways to do it. So that's, that's part one. And then we [02:27] always had this idea that like once we have a group of humans, we can use that to bootstrap a financial network on top of that. And so we're working a lot on how to make the crypto stack useful to everyday people. And so now we actually have the largest wallet in the world in terms of users. And we have a lot of people that [02:49] use it every day. And so that's super exciting. So you have WorldID, you have WorldChain, which is the blockchain that everything is running on top of. And we have WorldApp that makes all of that useful for people. And WorldID specifically, I think you will see integrated with many of the large platforms that you're already using today in the coming months, maybe, if we're good. And yeah. Yeah.

[03:15] So like Merge, or I guess like World and Merge, you like to go after somewhat of controversial spaces. And so biometrics has also perked some ears up. What are the biggest misconceptions that you're constantly facing? Well, I mean, the biggest one is, of course, around privacy. And I think what's so counterintuitive about what we're doing here is that we have designed a system that can be fundamentally anonymous. [03:45] is fundamentally anonymous and privacy-preserving. And [03:49] biometrics I think is the only way to actually solve the problem and I would need to go much deeper into why it is but I think it is the only way to solve the problem of human problem on the internet but the cool thing is with cryptography and and you know all the advancements that happened there over the last couple years you can build it fully on this and privacy preserving but I think that's like by far the biggest misconception but there's like other things around like do we do do we collect data and is that our business model and that's basically just all wrong like we um [04:19] We think that this is a network, so it has very strong network effects. And so we are very confident that if this keeps growing and keeps scaling and many platforms start integrating it, this will be very, very valuable. And we will be able to charge fees on the network and things like that. [04:37] But it's not the data of users, obviously. So that's another big misconception. Yeah, probably talk to you. So we're just coming off unwrapped and you have released some new features and products to the world with World. Can you share some of those highlights? What are you most excited about?

[04:58] Sure. So it was like very world app focused. [05:02] um so you know just taking [05:05] step back. The goal of this is to create the largest real human network on the internet because we think [05:11] that that's that will actually really matter, meaning [05:14] with the current AI progress and with everything that is happening, having some sense of... [05:22] real human interaction in every sense, like financially, socially, etc., will turn out to be like a really, really big deal. [05:29] much more than it seems today because I think like [05:33] 99.9% or something of all internet traffic will be AI driven. And so [05:39] That's, yeah, the [05:41] the high level. [05:43] We are working on Rolledout, which is, you know, so we're working on this network and we'll talk more about that. [05:49] World app is [05:51] our client for the network that we were building and we were just like trying to showcase how this network will be useful. [05:57] So-- [05:59] the network is a decentralized protocol that we're [06:03] working on, that we're trying to share with the world. The world app is actually our product, our Tools for Humanity. We're currently in Tools for Humanity office. And that's our product that we're working on. [06:15] and [06:17] what we released [06:19] Okay. [06:19] many new financial features. [06:22] So, you know, [06:23] paying with WorldCard, many new stablecoin integrations, like we've seen all the countries that we are operating with a lot of users.

[06:31] We have now 40 million users. [06:33] around the world, and so we're trying to support [06:35] different countries, different currencies, [06:38] That was a big one. [06:39] Um... [06:40] We released World Chat. [06:43] which is actually bringing all those things together. So it's a kind of a chat within a world app. And we have been super obsessed about the cryptography and the privacy side of that. So it's [06:55] that's extremely private but it also allows you to use all the other features of the network so the financial side [07:02] some of the world ID features so you can [07:06] You can use deep face, what we call that. So you can actually verify profile pictures of each other and, you know, and much more. So these were two of the big ones. [07:16] we actually showcased the Tinder integration, which was really funny, at least on stage. So meaning like, [07:23] within [07:24] Tinder and right now in Japan, it's already live. [07:28] you can prove both that you are in fact a real human, [07:33] you will be able to prove that these are authentic profile pictures that you're using. [07:38] And you can prove your age. So all things that really, really matter to-- [07:43] make this a [07:46] authentic dating experience and [07:49] you will see much more of that. So you will see like many big platforms integrating many of these different primitives that we're talking about. [07:56] somehow every week right now feels like Christmas. Like there's like so much [08:00] cool stuff and i'm really really excited about the company and i think the team is crushing it

[08:05] But I think the really, really cool stuff will come in Q1, and we were already working on that. So, and we're always one step already in the future. So the event was awesome. I think the team crushed it. I think the product was super cool. [08:18] But, [08:19] you know, so since tomorrow is even cooler. [08:21] You've also hit some really big growth numbers. Do you want to share some of those? [08:25] Yeah, you know, always there's always two sides to that. We're like on the one hand we did on the other side, we picked much bigger ones and we were aiming much higher. [08:36] We're now, I think, 37 million users total. 17 million of which are verified, so they use one of these orbs. [08:43] Um, [08:45] And [08:46] Yeah, there was like a lot of growth this year. [08:49] for that purpose, but [08:51] The mission is to [08:53] really get [08:54] to everyone. And so... [08:56] everything below hundreds of millions is not [08:59] doesn't mean much to us in some sense. Sorcery is brought to you by Brex, the financial stack trusted by more than 30,000 companies, including one in three venture-backed startups in the US. Nearly 40% of startups bail because they run out of cash. Brex is literally built to help founders avoid that. Unlike traditional banks that let your money sit idle, shipping away at it with fees, Brex is designed to help you spend smarter and move faster. Their all-in-one solution combines checking, treasury, [09:29] protection into one powerful account. You can send and receive money globally at lightning speeds, get 20 times the standard FDIC coverage through their partner banks, and even high yield from day one. With same day and even same hour liquidity, access your funds anytime. Companies like Scale AI, DoorDash, Service Titan, HIMSS, Anthropic, Flexport, Robinhood, and Plaid trust and use Brex.

[09:59] That's our ex.com slash sorcery. [10:29] R-C-E-R-Y. Yeah. I remember highlights from the presentation were hitting 16% APY on up to $1,000 in your account. That's insane. It is. Yeah. People go nuts over that. [10:45] 2 billion mini app users. That's pretty crazy. Uses, not users. Uses. 2 billion users would be a lot. Yeah, see then I would be happy. Soon. Yeah, soon. That's right, mini apps was a big one. And then the chat encryption I thought was really cool. I was talking to Shane Mack after and he was telling me about the rigorous integration process that he had to go through in order to make sure that the chat was encrypted and will be able to do that. [11:15] encrypted for 10 plus years so no one's leaking any information and he had to prove solving Shor's algorithm or something crazy like that yeah is this a typical process you put everyone through test their quantum skills well so we have like

[11:29] one of our core [11:32] So we have values and we have operating principles. And one of the operating principles is fail-safe privacy. Because... [11:40] in some like that that's the only way you could kill this company in [11:45] in a in a second like if like one engineer is doing something stupid and we would uh [11:52] leak user privacy, [11:54] Thank you. [11:55] basically in any form or shape, that would be so harmful to the project. And so like we are obsessed about privacy and [12:04] Yeah, and that's-- [12:05] And we have a very, very technical team. We have like [12:08] many photographers that [12:10] you know [12:11] are excited to be here. So I think we are on the very far extreme end on that dimension. [12:17] between all that what are the verticals you're most excited about for proof of human [12:22] Um. [12:24] So I think like just generally speaking, [12:27] My mental model of this is that I think [12:30] AI passed the Turing test, meaning it... [12:32] You know, it can talk to us on the internet, it can out-generate videos, it can do everything that we can do and it probably can [12:39] soon do it much better than we can do it. And so what that I think means [12:43] is that [12:45] truly every place on the internet where [12:48] humans are currently interacting with each other. [12:52] will start to break without [12:55] such a technology. [12:57] And not just interactions will break, but as a result, all the platforms

[13:02] that are built on human interaction will break. [13:05] And so, [13:07] First of all, that's quite bullish on the road. But second, [13:10] Um, [13:12] I think the initial use cases that I think make the most sense for this [13:16] are [13:17] Obviously social networks. [13:19] uh, [13:21] Dating is a big one. Gaming is probably going to be a big one. [13:26] and [13:30] just like other ones that you don't immediately think of. And it's kind of a secondary effect, but like, for example, [13:36] ticketing. [13:37] things like Ticketmasters and there's a lot of artists that are angry about the fact that there's bots buying all these tickets. [13:44] So, but that's roughly how I think about it. [13:46] and so i'm trying to spend time on each of these verticals and try to find the biggest [13:52] Integrations and also try to understand what are the you know, what are the things about our product that need to change? I [13:58] to make it [13:59] kind of more useful to them. And so that's like a, [14:03] big, big time spent from my side now. It's pretty crazy how fast the bots accrue. Like I remember when Elon first got into Twitter, now X, and he reduced all the bots. It like really lowered the amount of traffic on the platform and everything. And then all of a sudden it is completely picked back up. I have bots following my account all the time. I'm sure there's bots all over Reddit. And then that'll affect like model research and all that kind of stuff because it's all fake.

[14:33] So, [14:34] Good question. I think, first of all, [14:36] what i think the future will look like is much more of a [14:39] we will need to be able to distinguish [14:42] what is a human and what is an ai but i actually think we will probably want to have a lot of bots on our social platforms we just need to know that these are in fact bots but like once you really have [14:54] you know, super, super smart [14:57] AIs and some of them like have [15:01] whatever I find to you in a different ways and talk about different things that I think that's actually gonna be quite fun and quite cool and we will like that But we will need to know that am I currently talking to an AI system that is like trying to change my political view? Or am I talking to any to to a human that is maybe also in San Francisco and and so like me investing the time talking about [15:23] San Francisco city policy [15:26] is actually time-bound spend. [15:28] So that's kind of much more of how I think about the future, is like that I think [15:33] it's not about combating the bots because that will be a lost cause, but rather [15:38] every human on these platforms will need some version of a proof of human and and actually [15:44] prove that their content is authentic. [15:47] and then we will have all the rest [15:49] on the platforms. So there's that. And then to your point, I think like we're [15:54] Thank you. [15:54] just brought like we are at the bare beginning of how all of this is going to look like just because [15:59] once you have [16:01] true agentic systems and once you have

[16:04] these [16:05] very capable, organic systems also be open source. [16:09] things will go like quite exponential. So I think we are [16:13] still very, very early on the overall shape of how it would look like. [16:17] So I think [16:18] My personal prediction, I might be [16:21] wrong with this, but I think [16:23] platforms like X. [16:25] will [16:26] not be usable. [16:30] in like a matter of months. [16:32] maximum [16:34] two years. [16:35] if they don't come up with completely new technical [16:39] solutions. [16:41] I think what we're doing is very useful for that. Maybe there's other ideas that we are missing. [16:46] But at least currently I think it's the best idea that there is. I think a big question a lot of people [16:51] would love to hear and love to hear about is how you and Sam originally met, and then how it evolved to World, and then now to Merge. We originally met through a, you know, [17:01] a long set of [17:03] or random events, I think. [17:05] um so you know i grew up in germany [17:08] and [17:10] uh [17:11] I did a lot of things there. I did these science competitions and stuff like that. [17:17] Started a vertical farming company back then and but but nothing nothing like really [17:23] kind of baker or something that actually [17:25] But then I went into science. So I studied industrial engineering and physics. [17:30] and then within physics, [17:32] theoretical physics and within theoretical physics

[17:36] Um... [17:38] I got quite lucky because kind of next to my university, there was a Max Planck Institute. And [17:44] uh, [17:45] that just recently opened Max Planck, in Germany, is one of the best research institutions. [17:53] the director for theoretical physics at the institute started working on [17:57] deep learning and how you can apply deep learning to [18:00] all kinds of physical questions, but mostly quantum [18:03] quantum systems. Quantum systems are very, very hard to calculate. [18:08] especially if they're at scale and there's [18:11] So many interesting questions that come with that. [18:14] - Yes. [18:15] I started working on that topic and then I went to Caltech in Los Angeles and continued my research there. And in parallel, actually, I applied to YC120. [18:28] which, [18:29] was this conference that Sam hosted. [18:32] I think just once. And the idea was, [18:35] 100 of the smartest students in the world would mean 20 [18:39] top industry people or stuff like that. And you had to record a one minute video. [18:45] application video. I did that. [18:47] I got declined. [18:49] And I was like super angry at that. [18:52] And so I actually wanted to fly there and tell them that, but they were quite diligent about not sharing the address to protect against people like myself, I assume. I was never in the US before. [19:04] - Mm-hmm. - Nor ever Silicon Valley. Although I actually, like, I always wanted to come here. - Why?

[19:11] How did you get exposed to it growing up in Germany? [19:13] So I have this whole story about how I think how important that is and how people in Silicon Valley always underestimate how important it is that Silicon Valley has those. [19:23] um, [19:24] centralizing force on talent, but [19:27] I think the main thing I can remember [19:30] is while I was like super young, [19:34] for the purpose of that discussion. 11, 12, maybe. [19:39] maybe even slightly younger than that, there was an article. It was the title page of Der Spiegel, which is like the [19:47] one of the largest German newspapers, and it was talking about Silicon Valley. [19:52] And [19:54] I was like, [19:56] You know, I was obsessed with technology. I was like reading all these like technology [19:59] books. My dad was the opposite of that. My dad was like an economist. [20:04] And he [20:05] always made technology sound like [20:08] you know, not that serious and you should do more serious things. [20:11] in that story they were talking about this idea that there's silicon valley and there's all these people all these technologists they come together and they work on on things they think will be [20:20] change the future. Well, the story itself was actually very cynical. [20:25] I was too young to really realize that. But their Spiegel is like a little bit anti-tech. So I was like, [20:31] The underlying tone of the story was just making fun of Silicon Valley and how funny it is that all these people think that they can actually change the world with tech. But I read this and I was like, this is awesome. I want to go there. And so basically, since from that time on, I knew I would go to Silicon Valley one day.

[20:47] Told my parents, told my friends, everyone made fun of it. [20:50] That was kind of clear. [20:51] Hmm. [20:52] And then through YC and everything, you got exposed to Sam, you became working. Yeah, well, and then, yeah, right. And then like in high school, I watched all these YC videos and the interviews for VNode and all that stuff. So I [21:05] It wasn't Twitter and all these kinds of things. And then actually what happened is [21:11] I [21:12] I met Samu Burya, who you might know. [21:18] a really cool guy and he was at YC120. [21:22] I think... [21:23] Samo and Max, who was the other co-founder of World back then. [21:27] were connected and somehow that's how it came up. And so then I got an email from Max and Sam. [21:33] About world coin was like a two pager that both of them wrote so like both of them already were working on the project for like six six or seven months and [21:41] Yeah, then I [21:43] had many interviews with Max, and then I met Sam for an interview, and we kind of [21:48] hit it off um it was actually it was like a multi-month process and [21:53] I [21:54] The idea didn't make much sense to me initially, and so locked myself in the Caltech library book then, tried to make sense of it, still didn't make much sense. [22:02] So there was like many ups and downs. [22:05] almost declined the offer [22:07] I kind of already did. And then, but somehow we [22:10] we figured it out. And then, and I think, you know, now Sam and I work together for [22:16] what is it, six years? It's just become--

[22:18] better and better friends and now I think we're pretty close friends and [22:23] try to work on many things together. [22:25] And how do you [22:26] both manage [22:28] For him, OpenAI, you for World, well, I guess for him, OpenAI and World and all the other projects he's working on. And then for you, World, and then coming into Merge, how do you think you're going to lead these two companies? [22:40] So [22:42] The first question about world and how we work together on that is I think we have a pretty good cadence of [22:49] We usually try to do it once a week. Sometimes it doesn't work. But I try to send Sam kind of an in-depth update about anything that has happened in a week, and it's usually [23:00] Yes. [23:00] It depends on how much time we have in the company, but it's sometimes very in-depth. It's like 10 pages, all the major slides that have happened within a week. [23:09] and we just talk through everything. [23:11] And then [23:13] um, [23:14] my simple internal rule, and the cool thing is actually that like [23:18] Sam would probably [23:19] would have been fine either way. But my simple rules are like every major decision that I make, Sam should be involved in. So that is about like, [23:27] recruiting, strategy decisions, big spending decisions, etc. He has been super generous with his time on pretty much every important call that we had to do around fundraising or partnerships, integrations, all these kinds of things. So that's kind of how we're working is [23:42] Thank you. [23:43] trying to keep him up to date on the company operation, like once a week, and then get him involved in all the big strategic

[23:51] decisions and and work streams which could be financing partnerships hiring decisions mostly and then um for merge honestly i don't know yet i started waking up much earlier um and and like how's your sleep um [24:12] I kind of said goodbye to my social and personal life completely. I never really had it, honestly, since the startup world. But now it's gotten so extreme that I basically don't respond to messages anymore. And it's just like 14 hours a day, seven days a week. [24:26] all the time [24:27] Rebecca said, you're not going to make time for yourself until we reach singularity. I don't know the exact quote, but maybe, you know. Yeah, I mean, that's like, that's the most... [24:39] So to respond to the Merge question, first of all, I'm not running Merge in that sense. I'm not the CEO of Merge. But rather, we are [24:47] We are a group of co-founders. So it's currently, it's really structured like a research lab. It took us a [24:52] I think almost one and a half years to form the company. So we really spend a lot of time together. [24:56] And so that's really working. But back to my singularity question. [25:02] You know, like this is like very I just [25:04] if you take a step back, [25:07] the current moment in time is so crazy to me. It's just like this fact that... [25:12] We are probably in the last innings before Super Intelligence. [25:18] That means that if you're actually, if you are an individual that can make a meaningful impact on how that's going or like, you know, of course, the pessimistic take is you can't make any impact on anything. It's just like, this is just set on rails and super intelligence is going to happen and whatever happens after, who knows?

[25:35] You know, but that's... [25:36] That's at least not a decision I made for myself and for my life, I think. [25:39] This is like such a... [25:41] like infinitely leveraged moment in time where any contribution you can make might make actually, might make a meaningful impact. It might not at all, but it totally might. [25:50] and [25:52] looking back a couple hundred years, [25:54] we've never had that like you know there was like [25:57] humans, [25:58] thought about these kinds of problems, usually within a lifetime, basically nothing happened. [26:02] Like you maybe had like one scientific idea across the entire lifetime of a human and [26:10] much more so like literally nothing happened just you know you were just doing working on the field and you know uh and [26:19] Like right now, it's just a very different moment of time. It's where every day something new happens, every week there's a new big breakthrough. [26:27] And I think that will continue. And so like for me personally, I just make the decision that [26:32] Right now is not the time to... [26:35] Um... [26:37] Yeah. [26:38] spent much on the social life for all those things and I think [26:41] I'm just going to go completely hardcore for the coming years, at least five or so. [26:46] Um... [26:48] I might regret that, but so far it feels good. [26:50] It's painful, but just feels important.

[27:20] Founders ship faster on Deal. Set up payroll for any country in minutes, hire anyone anywhere, and get visas handled fast so you stay focused on scaling. Deal takes care of onboarding, HR, IT, ER, benefits, and compliance so your team can grow without borders. It's why more than 37,000 fast-growing companies trust Deal to move fast. Visit D-E-E-L dot com slash sorcery. That's D-E-E-L dot com slash S-O-U-R-C-E-R-Y. [27:50] couple people about you and in preparation of this. Okay. As I usually do, but it's nothing too scary. Um, [27:58] to describe you, there's two words. [28:01] People say that you're very loyal. [28:04] and you're very futuristic, and then you also are driven by like a really hard core set of values. Yep. So do you think that's accurate? And then what are your values? [28:15] I think that's broadly accurate. [28:17] I mean, [28:19] Futuristic is an interesting one because I think [28:22] We will see if I'm right. I try to think a lot about the future, obviously, but I think many, many people say give value. I think the interesting question is, are you correct with your predictions, which is the hard part. And who knows? Like I'm... [28:35] I'm way too young to know that. [28:37] But... [28:38] I'm loyal to very few people or like the bar is very high. [28:43] I think if I have a [28:44] close friendship i'm i'm very very loyal and i also just usually usually don't give them up so like i'm [28:50] quite restrictive on my

[28:53] social circle and [28:56] So it's very hard to get in there, I guess. [28:59] But... [29:00] But, you know, I think that's how a friendship should be. [29:04] that should be deep and meaningful and it should last many years. And so [29:08] Yeah, I would. [29:10] I would probably say that's true. [29:12] like my childhood was not always the easiest in that sense that I you know I was bullied and all these kinds of things but [29:20] I think the thing that I discovered for myself when I was around 15, 16, [29:25] is that I could always get forward with discipline. [29:28] And with like... [29:31] being perfectly truthful. [29:34] And [29:37] So like, you know, a big, big thing happened back then. We're like, my best friend died, who [29:42] Thank you. [29:43] You know, it was like [29:46] perfect person you could think of, honestly. [29:50] honest smart [29:53] and and that inspired me and so and so like [29:57] you know from that time on i i have this this long note which is probably at this point [30:03] It's probably 100 pages. [30:05] about all the things that I believe in, and kind of what are the core values and the principles that I try to live by. [30:12] And so... [30:13] So I'm probably erring way further on that than [30:17] normal people do. [30:19] And you keep on deciding to go after harder and harder things. [30:22] Yeah. [30:23] Why do you think it's important to try to go after hard things?

[30:26] I think that's-- [30:27] Isn't that the purpose of life a little bit? I think it is like meaning going after [30:34] trying to be the best version of yourself and trying to like take on as much responsibility as you possibly can [30:40] is how I would probably describe [30:43] the purpose of life, at least I've decided for myself. [30:47] And so you will probably keep me [30:49] I will keep doing that. [30:51] I'll just try to do as much as I possibly can, which I think is like going in the general right direction, and which I think matters for humanity. [30:58] Um. [30:59] and I will try to push myself as hard as I can to do that. Who are some idols or people that... [31:06] have mentored you along the way that you look up to you've mentioned sam a bunch and he's [31:10] part of your life, but who are some of those people? [31:13] Couple ones come to mind. I think like, or at least like people that really inspire me. [31:20] I will start with the people that I'm in touch with and then the people that just like broadly inspire me. [31:24] But I think Ben Horowitz, I think, is... [31:28] incredible and [31:30] Just like has such a good... [31:32] sends an opinion about [31:34] management and about like how to think about people and how to think about companies more broadly. And so I think truly incredible. We know at Kozla [31:44] and you know it was like he was one of the few people that i [31:49] you know, I was obsessed with when I was pretty young, like when I was in high school. [31:54] just because he took these like very, very big

[31:58] and he was like clearly [32:01] in the hard tech land, which I found super inspiring back then. [32:05] And for those two, actually, that's still the case, like meaning with both of them, I'm now in touch and I think both of them. [32:12] uh [32:14] I think I'm just like super wise and just know a lot of things, which is [32:19] Impressive. [32:20] Elon is of course like [32:23] incredible. Like, he has figured something out that no one else, I think, has yet. [32:28] So, quite inspired by that. I don't know him nor [32:32] we put it, right? [32:35] So, [32:36] I think these are the obvious ones. Well, I think we have some big news today. Can you please address the rumors? Which ones? [32:43] You don't know which one? No, I don't know which one. I want to talk about Merge Labs. Okay. [32:49] What happened? What happened? Well, I think we're announcing MorchLabs, which is a research lab [32:55] working on bridging artificial and biological intelligence to [33:01] you know, do a lot of things that we think will be important for the future. [33:03] And, uh, [33:05] Yeah, so the-- [33:07] We started working on it very recently. I'm really, really excited to [33:12] have found the most amazing co-founders [33:14] to pull this off. [33:15] Who's involved? [33:17] Um, [33:18] So it's kind of three camps coming together. So that's kind of how it happened. It was [33:23] Sam, Sandra, and me. [33:25] Oh, three of us. [33:27] working together since many years on Tools for Humanity.

[33:31] and other things. [33:33] then it's Mikhail Shapiro. [33:35] who's a professor at Caltech and has been working on BCIS for [33:39] 20 years. [33:40] And this, I think, one of the-- [33:42] most brilliant minds in that in that space. [33:44] And then it's Samson Tyson, who we're working on. [33:48] Forrest Neurotech, which was a focused research organization. They were working specifically in ultrasound for neuro. [33:55] But more broadly speaking, I think they're also just [33:57] really, really strong founders. [34:01] And with this announcement comes funding. So who did you raise from? [34:05] A large group of people, but but the biggest check was opening eye. [34:10] and [34:12] than many of our friends. You know, some of the friends that have been involved here [34:17] many, many individuals. It's like a broad, like it's a, like it's a, it's a large scale seed, but it is from that perspective, it is a classic seed where you try to, [34:26] bring in all the friends and [34:29] and people along the way. $250 million seed? $250 million, yeah. [34:33] Well, congrats. Thank you. [34:35] So what is the mission behind Merge and why is now the right time to go after it? The mission more broadly is bridging [34:42] artificial and biological intelligence, [34:44] to accomplish a lot of things that I think will be super important for the future. So like, uh, [34:50] agency, experience, freedom of people. And so it is about high bandwidth, [34:57] brain-computer interfaces, [34:59] and the relevant AI systems that we think will make such BCIs really, really useful. And so the two big statements to make here, on the one hand, it really starts like a research lab,

[35:10] because we think that [35:13] to get to high bandwidth. [35:16] BCIs that [35:17] are able to have a lot of coverage and cover large parts of the brain. [35:21] is at this point, [35:23] Very much a research question. [35:24] We are [35:25] like [35:26] You know, we spent a year [35:29] looking at everything and we [35:31] basically left convinced that none of the current approaches will actually get us there. [35:35] So that's why it's a research lab. [35:37] Um, [35:38] But the long term goal is clearly to build products that people love and want to use. And, you know, people like you and me would want to use. So [35:46] kind of the [35:47] broader and broader like if you now look at a 15-year [35:50] time horizon, I think. [35:52] It's going to be a couple of years of research, then a couple of years of medical applications and like hopefully bringing benefits to a lot of people that [36:00] really need such a technology. [36:02] And then eventually [36:04] get into kind of broad consumer applications. [36:07] Thank you. [36:08] And why is now the right time? [36:12] So first of all, it's a relatively large funding team, as I quickly mentioned. [36:17] just because in some sense it was like, [36:19] we call it like three camps are coming together, like the one camp is [36:24] kind of, well, Sam, Sandra, and me, which is Tools for Humanity, [36:29] And but just like a friend group, like we're not neuroscientists, obviously, but we were like super passionate and interested in the topic for a long time now. And so we spent some time on it.

[36:40] Mikhail, who I think is like one of the leading scientists in space and just like super incredible person. [36:47] Um, [36:48] and he is a professor at Caltech, and so has been working on these kinds of topics for over 20 years. [36:55] And then Sumner and Tyson from Forrest also have been thinking about BCIs for many, many years and have [37:03] build a lot of [37:04] cool things. So basically, like, [37:07] We're coming from [37:09] Yeah. [37:09] a diverse background [37:11] But [37:14] Why the time is now is like, I think on one hand, AI progress [37:18] is is just not stopping and it's continuing and so there have been [37:22] you know. [37:23] some new scientific [37:24] Uh, [37:25] results that are very promising that [37:28] Mikhail got very, very excited about. [37:30] For people who might not know what a BCI is or what BCI is, could you explain what that is? Oh yeah, sure. So it's a brain-computer interface. So something that directly inputs and outputs information into your brain. [37:44] connects to your neurons. So the most [37:47] famous example is obviously [37:50] Neuralink from Elon and what Neuralink is doing is having this robot that is inserting these threads and [37:57] into your brain and they're able to do a lot of really cool things with that so they can [38:02] uh, [38:03] you know, they can give someone that was not able to move [38:06] uh, [38:08] his arms for a very long time, now able to control the cursor and use a computer again, etc.

[38:15] Um... [38:16] That has been like a standard application. And there's more ideas, you know, [38:22] making blind people see again, or at least [38:27] approximately science and Max Hodak have been working on that, which I think is also super cool. So, but that's the general idea. The general idea is like BCIs are all kinds of systems that can directly interface with the brain. [38:39] And the long-term goal is that you get to [38:43] much higher bandwidth than what we can do with a keyboard. [38:46] Yeah, I guess that's cool. [38:48] People are saying this is gonna be the next major [38:51] technological advancement for computing after the smartphone. [38:55] Do you think that's... [38:57] Accurate? I'm like, you know, I'm much more bullish actually. So first of all, I think that's correct. That's why we're kind of starting to work on this. But I'm actually also quite excited about AR and VR. So I would say that like the [39:12] um, [39:14] if i could do a personal guess i think like the ar vr [39:19] step will be next. [39:21] And then I think BCIs and kind of [39:23] high bandwidth PTI specifically, [39:26] It's going to come after that. So I think it's going to be [39:29] you know [39:30] What current [39:31] smartphone, [39:33] situation then you will have ar vr and then i think you will have vci that would be my guess what is the technological differentiation for merge compared to neuralink and these other companies [39:45] a couple things without, you know, trying to get too technical way too quickly. But,

[39:52] for kind of the first and I think it's a quite important statement [39:56] is [39:57] we really just care about the mission. [39:59] And so we have multiple approaches that we will likely pursue. [40:03] So we are... [40:04] We have, I think, a brilliant team of many of the smartest scientists in the world working on this topic. And we have this long-term mission that we go after, and I think we will get there. [40:14] but [40:15] uh, [40:17] You know, we are not [40:19] obsessed about a particular approach or technology. [40:22] But... [40:24] The broad scheme that we think is is probably most promising right now with all the knowledge kind of all [40:31] all the results that we see right now and the current understanding we have. [40:34] of the space. [40:35] is that the combination of ultrasound and molecular sensors [40:39] is probably most promising. And then you can [40:43] both read and write. [40:45] and the limit with pretty much the entire brain. [40:49] Are there any applications you're excited about? [40:51] Just so you get a sense, the current BCIs that we're dealing with [40:57] they're all on the order of a couple thousand neurons. [41:00] that we can interact with. [41:03] and or, you know, actually neurons is not even [41:06] maybe precisely correct, it's channels. [41:09] So meaning this is what we're talking about here, if this really works. [41:12] is a [41:13] completely different level of BCI. [41:16] and [41:19] We don't even have the neuroscience for that yet. [41:21] With investment from OpenAI, how are you going to be working with them? Are there any collaborations there? Is this going to be seen as like an offshoot for research? How is this?

[41:31] implemented. [41:32] Well, first of all, we will find out given like, I think it's the first [41:41] you know, [41:42] This is like a first on many dimensions. [41:44] And so it's a direct investment from OpenAI, not the Venture Arm. And so it is an actual collaboration. [41:51] and [41:52] As we were talking through this investment, there were many different [41:56] dimensions that we thought could be interesting for collaboration. [42:01] I think there's a lot we can learn from OpenAI and how to do it at best. [42:04] And so, yeah, three major areas, I think, will be biology, chips, [42:09] AI systems. [42:10] Brain control. [42:12] Not brain control. [42:14] What do you think the biggest misconceptions are about BCI? I mean, a lot of people are going to use that in headlines. I think it's the obvious ones. It's probably like... [42:23] mind control, like what you just mentioned, I think is a [42:27] is a big one. [42:29] all concerns on privacy. Like, you know, if you actually [42:32] think this through and you have a system that is actually able to access your entire brain, will a company be able to know what you're thinking, etc. So I think that that's a big one. [42:43] The third one is probably just because [42:47] all the BCI systems that people are currently thinking of are so invasive. There's always this this idea that BCI is like, I don't know, like a quite [42:56] like maximally invasive and you have to go through big surgery and things like that. So that's that's another conception that I think I've sensed when I spoke to people.

[43:06] The good thing is I think all of them are very solvable. I think... [43:09] uh, [43:10] Again, it's going to be a big feat of engineering and there's going to be [43:14] a lot of fundamental questions that we and others in the space will have to answer. [43:18] But... [43:20] I'm very convinced we will be able to build systems that are [43:24] completely approvably just the way they're set up. [43:27] fully private [43:29] Don't share any information with any outside service. [43:32] are minimally invasive, are fully reversible so you can just [43:37] remove it if you don't want to, or upgradable, so you can get a better version if you want to. [43:42] I think these are all the main things that we will have to figure out in the coming years once we [43:47] kind of leave the research stage. [43:49] of that. [43:50] Alex, I have to ask, what is the most exciting thing in this office you're legally allowed to share? Legally is an important addition. [44:00] I think it's probably this one. This is like a... it's the control panel of a Soyuz space capsule. [44:08] that on a Saturday night I was like, I tried to figure out what I should put in my house. [44:17] - And this? - Yeah, this was like, there was like, and so I was like, I was looking at Sotheby's. I never, I was never bidding the Sotheby's auction. [44:25] But there was a Sotheby's space auction, so it just went rogue and they, you know, [44:30] I bought like [44:32] infinite amount of space stuff. [44:33] And so this is one of them. And this is like the control panel. And like this is actually quite famous for this one. Because that's a spinning like that's actually how to navigate it. It's like a spinning earth.

[44:45] And so now why this is disassembled is, [44:48] Yes. [44:48] our security team is like, you know, [44:51] Thank you. [44:52] As you can imagine, they're quite obsessed with these things. And so now we actually try to make everything functional. [44:57] -And so... -Oh, so it works. No, it's not working right now. But I think you should... Actually, over there, there's like the launch. The launch timer is on these books. -Mm-hmm. -On the Feynman books. [45:06] Um... [45:07] So but I think if you come come back again a couple months, I think everyone's gonna everything is gonna glow and maybe we are able to get the Thing to spin again. It's gonna go beep boop. Oh [45:18] I'm not sure it's going to be Beep. And this seems like it's like it's really informed the design of the orb. [45:25] I can see so much similarity. [45:27] Uh... okay. Good for you. I can't. Um... no, I mean, I bought it way later. I mean, the first... also the first orb was so ugly. Look at this. That's not ugly. No, the right one is very ugly. Like, the silver one, I think, is very ugly. I like the silver. You do? Yeah, I like shiny things. [45:51] Well, that was actually a theory theory or like people like shiny things. [45:54] And they don't? [45:59] I think it just didn't help with the, [46:01] make it seem trustworthy. It didn't help with that. I think it was just like, it seemed to sci-fi. Professional looking. Yeah. So we toned it down. [46:09] Yeah, I get that. Telling it down. That's also why we have all this wood and whatever. [46:13] Yeah. Yeah. Wow. OK, cool. So where did this come from?

[46:18] You got it at Sotheby's? I got it at Sotheby's and came out of a truck. Originally I don't know where it came from like well obviously it wasn't a Soyuz capsule so it comes from from Soviet Russia but uh [46:31] I'm not sure who owned it before that. [46:34] It might have been a museum. [46:36] I don't know. [46:37] - It's pretty cool. All right. This is legal. It's great. You have it in the office. - It's legal, yep. Yep. Yep. - And it's going to be working next week. We'll check back in. - No, it's going to take, it's going to take us a month, but... [46:48] You're working on it. What are your biggest concerns for the future? [46:51] My biggest concern for the future? This is a deep question. First of all, I'm actually quite optimistic. I think it's going to be like, I think we can make it an incredible, amazing, exciting future. And I think that's my personal default. [47:12] um [47:15] I mean, it would be something around AI misalignment and like something in that direction would probably be one. [47:24] But that also seems to trend in a much better direction. [47:27] over the last couple of years. [47:29] You're too optimistic. It's too hard of a question. It really is. It's actually, it is quite a hard question. I mean, there's obviously a lot of [47:36] potential downside scenarios. But I'm always just the next step is like, hey, we can probably solve it. We can probably fix it. So I'm just generally quite optimistic. Okay, fair. All right. I think that's it. That's a wrap. Hey, it's Molly. If you enjoy our interviews, check out our newsletter, sorcery.vc, where we deliver a once a week top deals and tech headlines email, and also go deeper on our podcast interviews. Subscribe to Sorcery today. And don't forget to subscribe to the

[48:06] Link in description to sign up.