Beren Millidge
Guest on Dwarkesh Podcast. 101 claims across 1 episode, Sep 2026.
Positions over time
Every pair of statements that was compared. Only a verified change counts as a change of mind; rewording, added nuance and different questions do not.
No two statements by this speaker have been compared yet.
Predictions, in their words
As stated, each with its moment. A date appears only where the speaker gave one.
“I think there's going to be like a long tail of like miscellaneous stuff which like some human can do which like will take the models like quite a while to do”
Expects: Models will probably take quite a while to handle the long tail of miscellaneous tasks that humans can perform. · no date given · not yet resolved
“like there's been a classic thing almost like marvex paradox right where like we see like you know we think of the ai being like if it can do this it's going to be amazing right like if it can solve these hard math problems if it can win a chess blah blah and then it solves these things and then it's like not that impactful obviously it's somewhat impactful but like not everything it's like if somehow that continues and like there's never like the true like spark of generalization that occurs i think that could lead to like the ai just being like extremely good at kind of everything that people like put into a benchmark to incorporate into an environment but like there's still some persistent like sim to real which is somehow blocking everything i think this is kind of unlikely i think we do actually see this kind of generalization even from RL in practice already but”
Expects: AI systems could become extremely good at benchmarked tasks while persistent simulation-to-reality problems limit their broader impact if they do not achieve a true spark of generalization. · no date given · not yet resolved
“like if it is just like ridiculously hard to like generalize meta learning plus like we don't solve container learning it's just like super hard and impossible like this would be my like default scenario in that case”
Expects: The speaker's default explanation for the absence of radically transformative superintelligence is that generalizing through meta-learning remains extremely hard and continual learning is not solved. · no date given · not yet resolved
“yeah i mean i agree it would be very i mean the only way for this to not happen is if like as you said somehow asymptotes like just before basically because we're already pretty close in my opinion to like where we'll start crossing like the human elo score and so we'll need to asymptote before that and like that's the only way you know in this scenario you post where like somehow we're sitting here in 2035 and like everything”
Expects: AI capability growth will not asymptote just before AI reaches human-level ELO; the discussed scenario would occur unless growth asymptotes before crossing that level. · no date given · not yet resolved
“everything is normal for this to happen i think yeah i mean the only other way is like there's like some dramatic like regulation on ai it's like this is kind of what i see is like the most likely way for this scenario to happen actually rather than the technical thing”
Expects: The discussed scenario might not occur because of dramatic AI regulation; the speaker considers this a possible and more likely reason than a technical limitation. · no date given · not yet resolved
“um coming back to more of x paradox there might be like a case of more of x paradox where like we think this kind of like autonomy and sort of like being like self-encapsulated so we can you know think of what we should do ourselves and then go do it and like have this loop is like super easy because we always do this and like obviously evolution needs to create creatures that can like survive on them by themselves like long period of time and like this just might be something that for some reason it's like really hard for the ai in the same way that like locomotion stuff is really hard but it's like math is super easy despite being super”
Expects: AI autonomy and self-directed action might be unexpectedly difficult for AI, despite humans routinely doing them. · no date given · not yet resolved
“suggest that that's you would yeah exactly i mean this is another possibility which like but i agree like there's no obvious evidence for this like in fact the fact that you know rl agents not being like super persistent and it's quite easy to do this it's kind of evidence against this but like this would be you know potentially like one of the reasons why like we just don't get this like immediate takeoff is like if this is hard”
Expects: An immediate AI takeoff may not occur, with difficulty achieving AI autonomy being one possible reason. · no date given · not yet resolved
“models and so it's like you can automate like an awful lot of this like prompt distribution gathering and like environment creation it's just like humans need to provide like increasingly few amounts of bits it's like the models get better”
Expects: As models improve, AI can automate much of prompt-distribution gathering and environment creation, reducing the amount of information humans need to provide. · no date given · not yet resolved
“Yeah, I mean like ultimately like if you have this like fully automated loop that is basically our side right like the AI deciding the data it's deciding the training that that is the loop but yeah I mean like it depends how much human information you need like at some point if you're just like I want traces that look like this you prompt that to the model the model will be able to like come up with like a pretty good approximation.”
Expects: When prompted to produce traces with specified characteristics, a model can generate a fairly good approximation. · no date given · not yet resolved
“can solve for the AI can always be still trying to solve it and so that would be the path to like go ahead of just like what the human AI research”
Expects: AI systems trained in environments with tasks beyond human ability could surpass the capabilities of human AI researchers. · no date given · not yet resolved
All 27 predictions
“Yeah, I mean, like presumably the models will be trained on like some combination of of all of these tasks and like some will be very easily verifiable some will be like LMS judge or like just ask the human like does this look reasonable and then you will say the hope would be that like these would all generalize to like these much sort of hard sort of more vague fuzzy kind of tasks and like it probably will to some extent whether it generalizes enough that like we could the loop can become like self-sealing without humans being in the loop at all it's like unclear”
Expects: Models will presumably be trained on a mixture of tasks, including easily verifiable tasks and tasks evaluated by an LLM or a human. · no date given · not yet resolved
“of all of these tasks and like some will be very easily verifiable some will be like LMS judge or like just ask the human like does this look reasonable and then you will say the hope would be that like these would all generalize to like these much sort of hard sort of more vague fuzzy kind of tasks and like it probably will to some extent whether it generalizes enough that like we could the loop can become like self-sealing without humans being in the loop at all it's like unclear”
Expects: Training across varied tasks will probably generalize to harder and less clearly specified tasks to some extent. · no date given · not yet resolved
“of all of these tasks and like some will be very easily verifiable some will be like LMS judge or like just ask the human like does this look reasonable and then you will say the hope would be that like these would all generalize to like these much sort of hard sort of more vague fuzzy kind of tasks and like it probably will to some extent whether it generalizes enough that like we could the loop can become like self-sealing without humans being in the loop at all it's like unclear”
Expects: Whether task generalization will be sufficient to make the AI research loop self-sealing without any human involvement is unclear. · no date given · not yet resolved
“we think Yeah like there is likely some transfer and sort of even just even if like finance is not specifically like the information is important for like RSI just the general like meta learning of like how to figure out what's important how to have taste how to like do long horizon work is potentially generalizable and like there's not that much RSI like data in the world as well like it's kind of hard to generate and like that requires a lot of effort so like if you can sort of advertise in this other data get some transfer from it you already have masses of computing masses of parameter space so like why not do that as well as like obviously”
Expects: Training on finance may transfer to recursive self-improvement by teaching general meta-learning skills, including identifying what matters, developing taste, and handling long-horizon work. · no date given · not yet resolved
“it right now you need like you know thousands and thousands of interactions with the humans and no human is going to sit there and like deal with this basically be in the loop of rl training and”
Expects: People will not manually participate in the thousands of interactions currently needed for reinforcement-learning training. · no date given · not yet resolved
“so like we kind of have to simulate that now to like get the samples you need but like obviously if sample efficiency improves a lot you'd expect learning from deployment to like become like a much bigger part of it this”
Expects: Substantially improved sample efficiency would make learning from deployment a much larger component of model training. · no date given · not yet resolved
“i think it will definitely like go through stages because because i mean this is assuming there's like one discontinuous event where suddenly we fix like”
Expects: The speaker expects continual model improvement to proceed in stages rather than through a single sudden discontinuous event. · no date given · not yet resolved
“i think it will definitely like go through stages because because i mean this is assuming there's like one discontinuous event where suddenly we fix like like weight updates continuously and like in practice i think it's much more likely to be like the cartridges and stuff allow you specialized in deployment then you generate traces you put that in your model like three months later you come up with a model which is better at this stuff you specialize it again you like consolidate it again and then eventually we'll just like make this loop faster and faster so instead of like every three months we release a model now it's like every week and then every like day and then every hour at which point we basically have always solved it”
Expects: AI learning from deployment experience will more likely develop through stages than through one discontinuous event that suddenly enables continuous weight updates. · no date given · not yet resolved
“like weight updates continuously and like in practice i think it's much more likely to be like the cartridges and stuff allow you specialized in deployment then you generate traces you put that in your model like three months later you come up with a model which is better at this stuff you specialize it again you like consolidate it again and then eventually we'll just like make this loop faster and faster so instead of like every three months we release a model now it's like every week and then every like day and then every hour at which point we basically have always solved it”
Expects: AI deployment learning will progress from specialized cartridges generating traces that are incorporated into a later model, followed by further specialization and consolidation. · no date given · not yet resolved
“like weight updates continuously and like in practice i think it's much more likely to be like the cartridges and stuff allow you specialized in deployment then you generate traces you put that in your model like three months later you come up with a model which is better at this stuff you specialize it again you like consolidate it again and then eventually we'll just like make this loop faster and faster so instead of like every three months we release a model now it's like every week and then every like day and then every hour at which point we basically have always solved it”
Expects: The speaker expects iterative model improvement to move from roughly quarterly updates to weekly, daily, and hourly updates, using deployment traces to train later models, until the process effectively solves the task continuously. · no date given · not yet resolved
“like weight updates continuously and like in practice i think it's much more likely to be like the cartridges and stuff allow you specialized in deployment then you generate traces you put that in your model like three months later you come up with a model which is better at this stuff you specialize it again you like consolidate it again and then eventually we'll just like make this loop faster and faster so instead of like every three months we release a model now it's like every week and then every like day and then every hour at which point we basically have always solved it”
Expects: The cycle of AI model improvement and release could accelerate from every three months to weekly, daily, and eventually hourly. · no date given · not yet resolved
“mean not necessarily i mean like maybe eventually if you have continuous learning you never train any model you just like just have a model and it keeps learning and like expanding right”
Expects: Continuous learning might eventually allow one model to keep learning and expanding, eliminating the need to train separate new models. · no date given · not yet resolved
“yeah i mean i would say maybe like for the like full generality maybe like three years but i think”
Expects: Models with the full generality of a remote worker may be available in about three years. · 11 Sep 2026 – 11 Sep 2029 · not yet resolved
“to Charlie's point we will end up with like a lot of people like making their organizations easier for the AIs to use and so you get like 80 90 percent of the way there before that like”
Expects: Organizations will adapt their systems for AI use, enabling AI to reach 80–90% of the described remote-worker capability before achieving full generality. · no date given · not yet resolved
“Yeah, I mean I think like this will mean that AI progress doesn't get bottlenecked on like AI research's ability to run like small experiments it gets bottlenecked on other things so”
Expects: AI progress will stop being bottlenecked by AI researchers’ ability to run small experiments and instead become bottlenecked by other factors. · no date given · not yet resolved
“and then if it can run like two or three experiments in a row without like crashing then like that is actually big update in like uplift”
Expects: The speaker's estimate of AI's uplift would substantially increase if AI successfully ran two or three consecutive experiments without crashing. · no date given · not yet resolved
“yeah i mean i kind of agree in like the the five-year range at least for like the stuff that like labs are focusing on but i think like there's going to be a long tail of stuff which like yeah i could theoretically go out and learn about but like no one has bothered to do it and like the computer isn't being allocated to that so that might take longer for like literally every single human expert”
Expects: AI could surpass human experts in areas that labs focus on within about five years, while this may take longer in less-studied areas that have not received comparable research or computing resources. · 11 Sep 2026 – 11 Sep 2031 · not yet resolved
Appearances
Claims per episode
- 11 Sep 2026AI researchers debate how close we are to recursive self-improvementDwarkesh Podcast · guest