Are short timelines actually bad?

Sam Altman recently posted the following:

I have seen very little serious discussion about whether short timelines are actually bad. This is surprising given that nearly everyone I talk to in the AI risk community seems to think that they are.

Of course, the question “was the founding of OpenAI net positive?” and “would it be good to accelerate capabilities in 2023?” are different questions. I’m leaning towards yes on the first and no on the second. I’ve listed arguments that factor into these questions below.

Reasons one might try to accelerate progress

Avoid/​delay a race with China. If the language model boom happened 10 years from now, China might be a bigger player. Global coordination seems harder than domestic coordination. A lot harder. Perhaps the U.S. will have to shake hands with China eventually, but the more time we have to experiment with powerful systems before then the better. That corresponds to time demonstrating dangers and iterating on solutions, which is way more valuable than “think about things in front of a white board” time.

Smooth out takeoff. FLOPS get cheaper over time. Data is accumulating. Architectures continue to improve. The longer it takes for companies to invest ungodly amounts of money, the greater the potential overhang. Shortening timelines in 2015 may have slowed takeoff, which again corresponds to more time of the type that matters most.

Keep the good guys in the lead. We’re lucky that the dominant AGI companies respect safety as much as they do. Sam Altman recently commented that “the bad case — and I think this is important to say — is, like, lights out for all of us.” I’m impressed that he said this given how bad this sort of thing could be for business—and this doesn’t seem like a PR move. AI x-risk isn’t really in the overton window yet.

The leading companies set an example. Maybe OpenAI’s hesitance to release GPT-4 has set a public expectation. It might now be easier to shame companies who don’t follow suit for being too fast and reckless.

Reasons to hit the breaks

There are lots of research agendas that don’t particularly depend on having powerful systems in front of us. Transparency research is the cat’s meow these days and has thus far been studied for relatively tiny models; in the extreme case, two-layer attention only transformers that I can run on my laptop. It takes time to lay foundations. In general, research progress is quite serial. Results build on previous results. It might take years before any of the junior conceptual researchers could get into an argument with Eleizer without mostly eliciting cached responses. The type of work that can be done in advance is also typically the type of work that requires the most serial time.

Language models are already sophisticated for empirical work to be useful. GPT-3-level models have made empirically grounded investigations of deception possible. Soon it might also make sense to say that a large language model has a goal or is situationally aware. If slowing down takeoff is important, we should hit the breaks. Takeoff has already begun.

We need time for field-building. The technical AI safety research community is growing very rapidly right now (28% per year according to this post). There are two cruxes here:

  1. How useful is this field building in expectation?

  2. Is this growth mostly driven by AGI nearness?

I’m pretty confused about what’s useful right now so I’ll skip the first question.

As for the second, I’ll guess that most of the growth is coming from increasing the number of AI safety research positions and university field building. The growth in AI safety jobs is probably contingent on the existence of maybe-useful empirical research. The release of GPT-3 may have been an important inflection point here. Funders may have been hesitant to help orgs scale before 2020, but I doubt that current AI safety job growth is affected much by capabilities progress.

My sense is that University field building has also been substantially affected by AGI-nearness – both in terms of a general sense of “shoot AI seems important” and the existence of tractable empirical problems. The Harvard/​MIT community builders found nerd-sniping to be much more effective than the more traditional appeals to having a positive impact on the world.

Attempting to put these pieces together

If I could slow down capability progress right now, would I? Yeah. Would I stop it completely for a solid 5 years? Hm… that one is not as obvious to me. Would I go back in time and prevent OpenAI and DeepMind from being founded? Probably not. I tentatively agree with Sam that short timelines + slow takeoff is the safest quadrant and that takeoff era has begun.

I mostly wrote this post because I think questions about whether capabilities progress is good are pretty complicated. It doesn’t seem like many people appreciate this and I wish Sam and Demis weren’t so villified by the x-risk community.