Thinking Machines’ Murati on AI’s Next Chapter

[00:00:00] Speaker 1: let's start with today and what you're working on you are building interaction models at thinking machines models that you say keep humans in the loop what exactly does that mean [00:00:14] Speaker 2: we've been working for the past year and a half on the foundations of building a frontier ai lab and with with the specific focus that we have and the interaction models were a first look at our concentrated bet towards human ai collaboration um and you know the the reason why we even started thinking machines is because we wanted to build the frontier ai lab that's really focused on um really human ai collaboration piece and that means a lot of things and there are specific research bets that go into that but we wanted to showcase um our work in one of the first bets through the interaction models and the interaction models are a new kind of model if you consider the types of models that we work with today they're very um turn-based and so you talk they talk then they go off and think um once you've given sort of uh prompt on what it is that you want to do and while they're thinking it's almost like they're deaf and blind they cannot perceive anything else about what's going on and then it's your turn and while you're talking they really cannot perceive anything about how you're talking it's not happening in real time and by contrast our interactions with each other are very rich there is a lot of information in our interactions when we are silent when we're thinking when we're interacting one another and so interaction models are able to capture all of these neurons they're not term-based they're more like time-based interaction where they're continuously taking in audio text video and continuously providing output now we cut this up in chunks of 20 milliseconds and this enables you to actually catch these things like interruptions and simultaneous speech and really create a rich high bandwidth interaction between humans and machines which we think is critical to actually enable um agency and uh and and enable people to be more in this loop [00:02:39] Speaker 1: as we advance ai further and further you're basically trying to build an agi lab from scratch you've got open ai anthropic google meta all with a significant head start all of them racing to build smarter models what is the bet that you're making that they aren't and what do you think they are underestimating [00:03:03] Speaker 2: so first of all i think um advancing the frontier of ai is incredibly positive sum and i think there is plenty of space for many different perspectives and ways of developing the technology and ways of developing the technology and i think um plurality is good um having a plurality of perspectives of ways of building technology different products is great for the world now this is a pretty hard thing to do so um i think that's why we don't actually have many players i think the barrier to entry is incredibly high um but in terms of creating something differentiated i do think there is plenty of room um i have always been very passionate about advancing the frontier of ai systems and i think that there is a potential for so much transformation for civilization that comes from that but it's not a predestined outcome the way that we go about building it and deploying the systems really matters and i think an area where there is uh very little work that's been done so far is um bringing the machine intelligence closer to where the knowledge is and so what i mean by that is there is one path of advancing frontier systems which is uh very autonomous and it doesn't rely too much on the messiness of reality or the um the the experience that humans have day to day and that's quite a fast way of advancing ai systems very autonomously autonomy is definitely a part of it and it's a very important part but i think a missing part where we haven't done much work is um really focusing on human intent of the messiness of interaction enabling people more and building um conceptually building frontier ai systems more like tools for thought where i mean i think the the most advanced ai systems are the the most incredible tools for thought that humanity can ever have and so how can this change the way that we think and where we're still thinking and uh but it's changing the nature of thought what we're thinking about um and and this part i don't think is actually this part is familiar to us you know um since the beginning of time like technologies have um deep technologies have changed what we think about like language writing uh numerals like imagine if you had to do multiplication with roman numerals so be miserable but you know we invented uh today's numerals and and this enabled a whole area of mathematics like a child could can do mathematics um very quickly and so it enabled this very tangible new ways of thoughts and i think this is the the ability to expand what we think about um and have new tangible things that we think about and this requires um but this this requires a very intentional uh research work and product work in this direction [00:06:40] Speaker 1: when i interviewed you it was early 2023 chat gpt had just changed everything you were cto of open ai a few folks i talked to said you basically ran the place just saying um when you left of to and and founded thinking machines were you running towards something or away from something [00:07:02] Speaker 2: um i most definitely was running towards something once i figured out what the thing was but i had an incredible i mean i had an incredible experience at open ai i was so incredibly lucky to work with some of the most dedicated and most talented people in the world and that's that's incredibly special and uh i'm very grateful for that experience um and you know eventually i had my own view of um and very strong view of how i think this technology ought to be developed and it's very rare to start something from scratch once you have uh developed such a strong perspective on it and that's that's a rare privilege to have and i think having a company like thinking machines gives us an opportunity to focus on where we have the highest conviction and build and orient the entire company around that conviction [00:08:14] Speaker 1: take me back to the board crisis of that year later that year you testified under oath that you were worried open ai was at catastrophic risk of falling apart you raised concerns about sam altman's leadership but in that moment you had to act looking back do you think you got that moment right [00:08:37] Speaker 2: it was a very intense time and it was a very complex situation and i'm sure many people have been in situations that they felt impossible where you have to make really hard calls and very quickly and i think there is a part of this complexity that wasn't to the world it appeared that it happened overnight but to me it was years of thinking about you know the mission the governance how you build a team that can build durable transformational technologies with that responsibility there was a lot of complexity to the organization that was there for years and we had thought about a lot this particular thing that happened this acute moment there was a lot of intensity intensity to it in time and you know when the board asked me for feedback i i shared feedback and i i stand by that and when they made the decision uh and asked me to step up as interim ceo i did that and um when i realized that their decision was like potentially catastrophic for the company and things would potentially fall apart i felt like i had to act very quickly and even though on the surface it just looked very chaotic i think at each point in time i i felt very clear about what i had to do uh because i the thing that i cared the most about was the mission the continuity of the mission and the people and the team and this is not like an abstract thing it's real people that i worked with for many many years and cared very deeply about um and all of this was about to you know potentially implode um so it was very clear to me what i had to do to create to create to to provide continuity stability and help bring send back restore that get the team in place to then deliver um gbd for oh and oh one um and so that was that was i think it was that principle that made it very clear to me what i had to do in retrospect um i i think that i would have paused more on understanding the transition plan and uh like of course with the benefit of hindsight there wasn't any transition plan and there wasn't uh much thought put into uh transparency bringing the team along providing a continuity and in retrospect i would have i would have paused more on that [00:11:28] Speaker 1: interesting helen toner testified former open ai board member mira was waiting to see which way the wind would blow and she didn't realize she was the wind what do you think would have happened if you [00:11:39] Speaker 2: didn't do what you did i think quite likely opening i would have imploded [00:11:48] Speaker 1: the case got thrown out on a technicality so the court never decided whether the leaders of open ai breached their mission how much does the character of the people building the most powerful technology in the [00:12:03] Speaker 2: world how much does that matter whose hand is on the dial i think the integrity character values of the people matter a lot because there are a lot of decisions micro decisions you have to make every day and you have to trust your team that they're making the right calls um and i do think there's plenty of focus on that and that's great where there is less focus is sort of like the institutional design decision making transparency um increasing the level of agency uh that people have to actually decide for themselves so it doesn't have to be like an authority that says you know this is safe this is good stamp of approval but each person can sort of wait for themselves um what works for them and i think the way that we're going about building ai systems today is just very concentrated and it's a part of the the legacy way because you know we didn't have real world uh data and experience we the only way to to advance ai was sort of in silo and in a vacuum but i think yeah today is it's a different time to build where we can actually learn a lot from the capabilities tensions limitations in the real world and use that data and information to actually steer the direction of research and this enables people to actually think for themselves like often people will ask me about you know chat gpt and how that the impact of all of that and of course it's magnificent from a technology perspective but i think the most important aspect of our impact of child gpt was bringing ai in the public consciousness and people every everyone kind of understanding what it is by interacting with it versus being told what ai can do and what [00:14:07] Speaker 1: what the capabilities and limitations are what do you think the aim this moment demands of a leader and should we trust the the leaders in power right now should we trust saml [00:14:25] Speaker 2: i think that everyone should be everyone should have the tools and information to be able to make these decisions for themselves and ideally like you know the structure of governance uh in decision making should not hinge on one person there should be checks and balances um i think it's very important who is working on the systems and you know the character of people but uh i i do think that the conversation gets too wrapped up there and not enough thinking more broadly about checks and balances and system because even people that are well intent well intentioned they can make mistakes and uh they can misestimate the the uh consequences of making a call and so morality is not everything you have to think about actual you know decision making structures and transparency and governance and these are all complicated things that you have to do and that you have to do and that you have to make a lot of things and to get as many people involved in these things as possible you actually have to share the knowledge you have to share uh tools and this is part of the reason why we've actually um taken a more open approach with our lab so that was my next question which is how is everything that you've learned [00:15:49] Speaker 1: shaping what you're building at thinking machines and how you're building and the culture that you want to create [00:15:59] Speaker 2: yeah i think i i have very high conviction that the way to continue building frontier ai systems is to bring people along and to have humans in the loop like actually having humans in the loop doesn't quite describe it because it sounds like a checkpoint where we're signing off something and then you're good to go it's more like creating systems that are not just like autonomously advancing and leaving civilization behind but are more like a tandem bike where you know you have like both both people are pedaling but you know when you're going up a hill maybe whoever is stronger is pedaling harder but both hands are on the are on the wheel and that's quite important because that's a different system it's a system designed for collaboration um and and that's that's what we're trying to build with thinking machines and i think it's quite differentiated and um and and i hope that it will increase the level of agency um that people have and also it will help us steer the research direction towards creating outputs that are more value aligned and so you also get alignment as a result of this approach in in addition to usefulness like actually creating technologies that are useful in the real world [00:17:36] Speaker 1: you've hired a lot of top talent we've reports of nine figure deals there have also been some high profile departures how would you describe the war for talent how brutal is it and what should we read into these exits [00:17:54] Speaker 2: it's definitely a big part of building a frontier ai lab having the right people and that's you know people that have the competence to do it but also are aligned with your overall mission conviction what you're trying to achieve i yeah i wouldn't call it a war because it might mean that the the highest bidder wins and and i think for the most part people that are working in this field are very you know they they generally care about advancing the field now it is an absolutely crazy time and a lot of it is unprecedented it's incredibly intense to build the frontier ai lab from scratch during such competitive times and so that leads to you know a lot of uncertainty and predictability some some of the volatility that we've seen um but yeah i think when you're trying to compress progress in such a short amount of time things that happen in other startups companies over the course of 10 years five years both good and bad are going to happen in the course of a few months and in a course of a year and i think part of um part of what we're seeing is contextualized in that just compressing uh the the amount of progress that's happening and whenever you have something good something bad will come with it and so you have to balance the two but people live for different reasons you know maybe uh the the the their passion evolved for what they want to work on um i think the the high numbers and the compensation numbers they capture the imagination of people because obviously they're very big um but i do think that you know some of the most sought after people uh that's that's not that's not the main story and i think there's a lot uh that goes on here and it is very intense to build a company from scratch let alone a frontier ai lab um so that's part of you know it's part of the challenges and suffering of building a company and doing anything zero to one daria abade has [00:20:26] Speaker 1: predicted mass you know white collar job loss sam though recently walked back some of his predictions about a you know job jobs apocalypse where do you think the industry is being too pessimistic and too optimistic like what is mira marati's vision for the future so you know predicting [00:20:49] Speaker 2: sort of a dystopia or utopia uh to me feels feels very very simplified because the truth is we actually have a lot of agency in how we build this technology in the tools that we're building how we're deploying it and and so it's not a predestined outcome there are certainly those risks we all understand the potential for greatness that comes with building frontier eye systems and that's why we're working on them and i think there's been a lot of talk around the uncertainty around the downsides and i i agree with a lot of these risks and perhaps where i might disagree or take a different path is that i think we have a lot of agency like this period of time where both humans and ai systems have their hands on the wheel and we can collaborate it's very important time to get right and the more that we can reduce the discontinuity that comes from new capabilities and huge change in capabilities the better um and and this is why we took this approach with thinking machines and why the company exists so you say the goal is to keep [00:22:13] Speaker 1: humans in the loop but is there ever a point where humans don't need to be in the loop and then what happens [00:22:19] Speaker 2: it is it is it is possible um it's it's more about aligning at the point it's more about aligning intent and aligning values but if you if you sort of reduce progress to removing humans from this loop of development already now then i see very little future possibilities that we can get this right in the future when ai systems are even more capable and so i think the more likely path and more durable path to to advance ai in a way that enables and advances civilization is by keeping humans in the loop for as long as we can and really making sure that this period is actually very meaningful [00:23:14] Speaker 1: you've raised a lot of money there's great expectations if anything's clear from you know the last few months in in ai news uh there are some you know it's a cutthroat competition out there do you think you have that killer instinct to take on open ai and anthropic and meta and google and china and x [00:23:34] Speaker 2: like it's a competitive world it is extremely competitive and uh yeah we did raise a lot of money we're we're proud of that but that's that's not any big accomplishment we didn't set out to break some sort of record um it's really what you do with that and uh and and it's it we're not a normal company and so we do need a lot of capital uh in order to build the infrastructure the foundations of science and all that's required to actually have a claim at building frontier ai systems that are differentiated um and so i think that part in itself um is not unique and yeah in terms of killer instincts i say that's not really what motivates me um to you know uh uh like i think there's actually a lot of room we're building we're building a technology that has you know infinite uh potential and i think that it's going to be hard to capture all of that potential and what motivates me is really capturing part of that potential creating uh useful things in the world bringing the technology into the world in a way that's uh useful and increases human agency and advances our civilization i think you know there is there is healthy competition and competition is good it it actually like creates better products and better technologies for people it raises the bar in many ways um and and you know i i respect a lot of the companies that that are doing this uh but my my motivation is not necessarily day to day when i wake up in the morning i am not thinking about how to kill the competitor [00:25:36] Speaker 1: the other companies that are doing this um you said you plan to release a preview later this year how far along are you what will we see with all of these other companies going public do you think they'll be [00:25:52] Speaker 2: distracted and you have a little more room to run um we're we're on our own path of course the competition is fierce and we have to move quickly but it's important to balance that with you know durable long-term progress i care a lot about the decisions that we make today being good for our long-term success and and that's something that i really try to instill in the team day to day there's a lot of details here with how you build the team how you build the infrastructure and really resisting the short-term pressures um and and in in in some cases you do have to kind of embrace the pressure but yeah i would say for us this is just a first step um and in showcasing the interaction models and making it more concrete what it is that we think about and care about um and i expect the next few months and uh to to show increased capabilities for us to show increased capabilities on the model side um more products in this direction and uh and make it more obvious like really letting the technology [00:27:11] Speaker 1: and the products speak for themselves so a thinking machine succeeds beyond your wildest expectations what what exists in five years that doesn't exist today [00:27:24] Speaker 2: it is extremely difficult to predict the future but i would say that the most important thing would be to create a future where regardless of how how many of hours of work we do day to day or week to week we feel a sense of agency a sense of dignity and um possibility about the future and i hope that we'll continue to advance the capabilities that will enable a lot of of progress and we'll do it in this way that um will keep will keep us hopeful and also with a sense of possibility that we can drive about the future well here's to less impossibility and more possibility thank you