John Schulman
Guest on Dwarkesh Podcast. 121 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.
“Humans have a lot of advantages over models now and each time a new model comes out, it'll sort of, it'll catch up in some of these areas.”
Expects: Each new AI model is expected to catch up with humans in some of the areas where humans currently have advantages over AI models. · no date given · not yet resolved
“I would imagine a future where AI is doing a lot of analysis and theory building, like spending a comparable amount of compute to the amount that you're spending on the experiments themselves, doing various kinds of analysis and building a theory around what we've seen so far.”
Expects: AI will eventually perform substantial research analysis and theory-building, using compute comparable to the compute spent on experiments. · no date given · not yet resolved
“I think we'll probably do some combination of learning from human feedback to absorb like the researchers taste and just like creating a lot of practice environments which involve like doing multi-step research projects.”
Expects: The first AI systems capable of automating AI research and development will probably combine human feedback that captures researchers' preferences with many practice environments for multi-step research projects. · no date given · not yet resolved
“I would expect the models to keep getting bigger just because people are scaling up compute and the GPUs are getting bigger.”
Expects: Models will probably continue to grow in size as available compute scales up and GPUs become larger. · no date given · not yet resolved
“So that cycle just might keep going and it's hard to predict how many times it's going to repeat.”
Expects: The cycle of initial excitement about new AI models followed by disappointment might continue; the number of repetitions is difficult to predict. · no date given · not yet resolved
“I mean, there is a possibility that there will be company specific models that it'll be possible to learn from deployment and have a company continually improving its own model.”
Expects: Companies may use company-specific models that learn from deployment and are continually improved by the company. · no date given · not yet resolved
“And such a system could be provided by the current oligopoly of model providers or some other currently smaller company.”
Expects: Company-specific, deployment-learning models could be supplied either by the current oligopoly of model providers or by a currently smaller company. · no date given · not yet resolved
“I mean, if the models were good enough at learning in context, then in theory, you wouldn't need to train them on finance. They would just be able to figure out, read all the books on the fly and figure out how to do everything in the appropriate jurisdiction.”
Expects: Sufficiently capable in-context learning could let models avoid finance-specific training by learning from books on the fly and adapting their work to the relevant jurisdiction. · no date given · not yet resolved
“And if you use some kind of superficial signal, like did they accept the code, the edit, that might get reward hacked in some way.”
Expects: A superficial reward signal, such as whether a user accepts a code edit, might be reward-hacked. · no date given · not yet resolved
“but then there might be some like medium length regime where they're less sample efficient because humans can. do some kind of weight update more efficiently than models.”
Expects: In some medium-length learning regimes, humans may update their weights more sample-efficiently than models. · no date given · not yet resolved
All 27 predictions
“but it's and like sparsity has definitely increased a bit, but it's not clear that it's going to keep increasing without bound.”
Expects: Whether model sparsity will keep increasing without limit; it is uncertain whether this will happen. · no date given · not yet resolved
“So that's something AI is obviously going to be extremely good at if trained for it.”
Expects: AI will be extremely good at solving hard search problems if trained for those tasks. · no date given · not yet resolved
“that maybe, yeah, we'll have some version of this in a year or so that's like, okay, and it will be able to do, maybe we'll have that form factor and it'll be able to do some things really well, some things not so well, and things will be improving from there.”
Expects: A version of the described AI remote-worker capability will exist in about a year, perform some tasks very well and others less well, and improve thereafter. · 11 Sep 2027 – 11 Sep 2027 · not yet resolved
“Like we might already be past that in some like types of work.”
Expects: AI may already have achieved the proposed productivity uplift in some types of work. · no date given · not yet resolved
“which models are really good at. Maybe for things that involve like 3D and like spatial stuff and physical stuff, I think that that will take a little longer.”
Expects: AI will take longer to become capable at tasks involving 3D, spatial, or physical work. · no date given · not yet resolved
“that even if the model can write way more code than a person it doesn't make you like a hundred times more productive but yes so maybe maybe there are just more of these cycles than we would than we would expect”
Expects: The cycle of excitement about new AI models followed by disappointment may repeat more times than people expect. · no date given · not yet resolved
“job for humans uh or the the role for humans that will last the longest is like defining the objective uh and like deciding what we actually want um so like in that vein”
Expects: As AI capabilities advance, defining objectives and deciding what people want will remain the human role that lasts longest. · no date given · not yet resolved
“something like deciding what how is it like the AI the assistants should behave or what it means to be helpful or what's like the objective when we're doing our all from human feedback is one such thing and then like then later like defining like constitutions and model specs is another one and I think even if the AIs can do all the technical work we'll have to still do a lot of that and decide what we actually want.”
Expects: Humans will continue to define desired AI behavior and objectives—including helpfulness, human-feedback objectives, constitutions, and model specifications—even if AI systems can perform all technical work. · no date given · not yet resolved
“Yeah, alignment is sort of the answer, but it's also alignment itself can be kind of decomposed into into like specification of the objective or figuring out what the right objective should be. And then. like actually like achieving or optimizing the objective you've defined and I think the first one is not going to go any way anytime soon”
Expects: The need for humans to determine the right objective in AI alignment will persist for the foreseeable future. · no date given · not yet resolved
“But I think that'll change the game a bit.”
Expects: The speaker expects company-specific models that learn from deployment to change the competitive landscape somewhat. · no date given · not yet resolved
“People will in practice do some combination of those two things and just each iteration like patch whatever seems to be most broken in the last iteration. So like researchers will be using the AIs a lot and will notice that they have some consistent weaknesses and then those things will either be patched by like collecting human feedback or like creating environment.”
Expects: AI developers will likely combine human feedback and practice environments, addressing the most apparent weakness in each successive iteration. · no date given · not yet resolved
“People will in practice do some combination of those two things and just each iteration like patch whatever seems to be most broken in the last iteration. So like researchers will be using the AIs a lot and will notice that they have some consistent weaknesses and then those things will either be patched by like collecting human feedback or like creating environment. the environments”
Expects: As researchers use AI systems, they will identify recurring weaknesses and address them through human feedback or purpose-built training environments. · no date given · not yet resolved
“bigger but i would say exactly how much they get bigger depends a bit on the scaling laws in non-obvious ways so one thing is that i think like data efficiency is going to be a bigger driver than compute efficiency of like the exact architectures people use now that we're getting to the regime where we're sort of running low on like high quality pre-training data so that might affect how sparse you want to make the model”
Expects: As high-quality pre-training data becomes scarce, data efficiency may matter more than compute efficiency in determining the extent of model growth, which depends in non-obvious ways on scaling laws. · no date given · not yet resolved
“There might be some kind of sweet spot.”
Expects: There may be an optimal level of model sparsity, rather than model sparsity increasing indefinitely. · no date given · not yet resolved
“I would say like two years.”
Expects: AI researchers may achieve a 10x productivity uplift in about two years. · 11 Sep 2026 – 11 Sep 2028 · not yet resolved
“I would say like three or four years.”
Expects: The speaker estimates that AI will reach the level of capability being discussed in three to four years. · 11 Sep 2029 – 11 Sep 2030 · not yet resolved
“So especially if it's not like, yeah, if it's like mechanical engineering or something and it's not. Not getting like the most attention right now that might take a little longer”
Expects: AI may take longer to master fields such as mechanical engineering when those fields are not receiving much attention. · no date given · not yet resolved
Appearances
Claims per episode
- 11 Sep 2026AI researchers debate how close we are to recursive self-improvementDwarkesh Podcast