What is the probability that the U.S. AI industry (including OpenAI, Anthropic, Microsoft, Google, and others) is in a financial bubble — as determined by multiple reliable sources such as The Wall Street Journal, the Financial Times, or The Economist — that will pop before January 1, 2031?
I put 30% on this possiblility, maybe 35%. I don’t have much more to say than “time horizons!”, “look how useful they’re becoming in my dayjob & personal life!”, “look at the qualitative improvement over the last six years”, “we only need to automate machine learning research, which isn’t the hardest thing to automate”.
Worlds in which we get a bubble pop are worlds in which we don’t get a software intelligence explosion, and in which either useful products come too late for the investment to sustain itself or there’s not really much many useful products after what we already have. (This is tied in with “are we getting TAI through the things LLMs make us/are able to do, without fundamental insights”.
I haven’t done the sums myself, but do we know for sure that they can’t make money without being all that useful, so long as a lot of people interact with them everyday?
Is Facebook “useful”? Not THAT much. Do people pay for it? No, it’s free. Instagram is even less useful than Facebook which at least used to actually be good for organizing parties and pub nights. Does META make money? Yes. Does equally useless TikTok make money? I presume so, yes. I think tech companies are pretty expert in monetizing things that have no user fee, and aren’t that helpful at work. There’s already a massive user base for Chat-GPT etc. Maybe they can monetize it even without it being THAT useful. Or maybe the sums just don’t work out for that, I’m not sure. But clearly the market thinks they will make money in expectation. That’s a boring reason for rejecting “it’s a bubble” claims and bubbles do happen, but beating the market in pricing shares genuinely is quite difficult I suspect.
Of course, there could also be a bubble even if SOME AI companies make a lot of money. That’s what happened with the Dot.com bubble.
This is an important point to consider. OpenAI is indeed exploring how to put ads on ChatGPT.
My main source of skepticism about this is that the marginal revenue from an online ad is extremely low, but that’s fine because the marginal cost of serving a webpage or loading a photo in an app or whatever is also extremely low. I don’t have a good sense of the actual numbers here, but since a GPT-5 query is considerably more expensive than serving a webpage, this could be a problem. (Also, that’s just the marginal cost. OpenAI, like other companies, also has to amortize all its fixed costs over all its sales, whether they’re ad sales or sales directly to consumers.)
It’s been rumoured/reported (not sure which) that OpenAI is planning to get ChatGPT to sell things to you directly. So, if you ask, “Hey, ChatGPT, what is the healthiest type of soda?”, it will respond, “Why, a nice refreshing Coca‑Cola® Zero Sugar of course!” This seems horrible. That would probably drive some people off the platform, but, who knows, it might be a net financial gain.
There are other “useless” ways companies like OpenAI could try to drive usage and try to monetize either via ads or paid subscriptions. Maybe if OpenAI leaned heavily into the whole AI “boyfriends/girlfriends” thing that would somehow pay off — I’m skeptical, but we’ve got to consider all the possibilities here.
What do you make of the fact that METR’s time horizon graph and METR’s study on AI coding assistants point in opposite directions? The graph says: exponential progress! Superhuman coders! AGI soon! Singularity! The study says: overhyped product category, useless tool, tricks people into thinking it helps them when it actually hurts them.
Yep, I wouldn’t have predicted that. I guess the standard retort is: Worst case! Existing large codebase! Experienced developers!
I know that there’s software tools I use >once a week that wouldn’t have existed without AI models. They’re not very complicated, but they’d’ve been annoying to code up myself, and I wouldn’t have done it. I wonder if there’s a slowdown in less harsh scenarios, but it’s probably not worth the value of information of running such a study.
I dunno. I’ve done a bunch ofcalibrationpractice[1], this feels like a 30%, I’m calling 30%. My probability went up recently, mostly because some subjectively judged capabilities that I was expecting didn’t start showing up.
My metaculus calibration around 30% isn’t great, I’m overconfident there, I’m trying to keep that in mind. My fatebook is slightly overconfident in that range, and who can tell with Manifold.
There’s a longer discussion of that oft-discussed METR time horizons graph that warrants a post of its own.
My problem with how people interpret the graph is that people slip quickly and wordlessly from step to step in a logical chain of inferences that I don’t think can be justified. The chain of inferences is something like:
AI model performance on a set of very limited benchmark tasks → AI model performance on software engineering in general → AI model performance on everything humans do
I put 30% on this possiblility, maybe 35%. I don’t have much more to say than “time horizons!”, “look how useful they’re becoming in my dayjob & personal life!”, “look at the qualitative improvement over the last six years”, “we only need to automate machine learning research, which isn’t the hardest thing to automate”.
Worlds in which we get a bubble pop are worlds in which we don’t get a software intelligence explosion, and in which either useful products come too late for the investment to sustain itself or there’s not really much many useful products after what we already have. (This is tied in with “are we getting TAI through the things LLMs make us/are able to do, without fundamental insights”.
I haven’t done the sums myself, but do we know for sure that they can’t make money without being all that useful, so long as a lot of people interact with them everyday?
Is Facebook “useful”? Not THAT much. Do people pay for it? No, it’s free. Instagram is even less useful than Facebook which at least used to actually be good for organizing parties and pub nights. Does META make money? Yes. Does equally useless TikTok make money? I presume so, yes. I think tech companies are pretty expert in monetizing things that have no user fee, and aren’t that helpful at work. There’s already a massive user base for Chat-GPT etc. Maybe they can monetize it even without it being THAT useful. Or maybe the sums just don’t work out for that, I’m not sure. But clearly the market thinks they will make money in expectation. That’s a boring reason for rejecting “it’s a bubble” claims and bubbles do happen, but beating the market in pricing shares genuinely is quite difficult I suspect.
Of course, there could also be a bubble even if SOME AI companies make a lot of money. That’s what happened with the Dot.com bubble.
This is an important point to consider. OpenAI is indeed exploring how to put ads on ChatGPT.
My main source of skepticism about this is that the marginal revenue from an online ad is extremely low, but that’s fine because the marginal cost of serving a webpage or loading a photo in an app or whatever is also extremely low. I don’t have a good sense of the actual numbers here, but since a GPT-5 query is considerably more expensive than serving a webpage, this could be a problem. (Also, that’s just the marginal cost. OpenAI, like other companies, also has to amortize all its fixed costs over all its sales, whether they’re ad sales or sales directly to consumers.)
It’s been rumoured/reported (not sure which) that OpenAI is planning to get ChatGPT to sell things to you directly. So, if you ask, “Hey, ChatGPT, what is the healthiest type of soda?”, it will respond, “Why, a nice refreshing Coca‑Cola® Zero Sugar of course!” This seems horrible. That would probably drive some people off the platform, but, who knows, it might be a net financial gain.
There are other “useless” ways companies like OpenAI could try to drive usage and try to monetize either via ads or paid subscriptions. Maybe if OpenAI leaned heavily into the whole AI “boyfriends/girlfriends” thing that would somehow pay off — I’m skeptical, but we’ve got to consider all the possibilities here.
What do you make of the fact that METR’s time horizon graph and METR’s study on AI coding assistants point in opposite directions? The graph says: exponential progress! Superhuman coders! AGI soon! Singularity! The study says: overhyped product category, useless tool, tricks people into thinking it helps them when it actually hurts them.
Pretty interesting, no?
Yep, I wouldn’t have predicted that. I guess the standard retort is: Worst case! Existing large codebase! Experienced developers!
I know that there’s software tools I use >once a week that wouldn’t have existed without AI models. They’re not very complicated, but they’d’ve been annoying to code up myself, and I wouldn’t have done it. I wonder if there’s a slowdown in less harsh scenarios, but it’s probably not worth the value of information of running such a study.
I dunno. I’ve done a bunch of calibration practice[1], this feels like a 30%, I’m calling 30%. My probability went up recently, mostly because some subjectively judged capabilities that I was expecting didn’t start showing up.
My metaculus calibration around 30% isn’t great, I’m overconfident there, I’m trying to keep that in mind. My fatebook is slightly overconfident in that range, and who can tell with Manifold.
There’s a longer discussion of that oft-discussed METR time horizons graph that warrants a post of its own.
My problem with how people interpret the graph is that people slip quickly and wordlessly from step to step in a logical chain of inferences that I don’t think can be justified. The chain of inferences is something like:
AI model performance on a set of very limited benchmark tasks → AI model performance on software engineering in general → AI model performance on everything humans do
I don’t think these inferences are justifiable.