In recent months, the CEOs of leading AI companies have grown increasingly confident about rapid progress:
* OpenAI's Sam Altman: Shifted from saying in November "the rate of progress continues" to declaring in January "we are now confident we know how to build AGI"
* Anthropic's Dario Amodei: Stated in January "I'm more confident than I've ever been that we're close to powerful capabilities... in the next 2-3 years"
* Google DeepMind's Demis Hassabis: Changed from "as soon as 10 years" in autumn to "probably three to five years away" by January.
What explains the shift? Is it just hype? Or could we really have Artificial General Intelligence (AGI)[1] by 2028?
In this article, I look at what's driven recent progress, estimate how far those drivers can continue, and explain why they're likely to continue for at least four more years.
In particular, while in 2024 progress in LLM chatbots seemed to slow, a new approach started to work: teaching the models to reason using reinforcement learning.
In just a year, this let them surpass human PhDs at answering difficult scientific reasoning questions, and achieve expert-level performance on one-hour coding tasks.
We don't know how capable AGI will become, but extrapolating the recent rate of progress suggests that, by 2028, we could reach AI models with beyond-human reasoning abilities, expert-level knowledge in every domain, and that can autonomously complete multi-week projects, and progress would likely continue from there.
On this set of software engineering & computer use tasks, in 2020 AI was only able to do tasks that would typically take a human expert a couple of seconds. By 2024, that had risen to almost an hour. If the trend continues, by 2028 it'll reach several weeks.
No longer mere chatbots, these 'agent' models might soon satisfy many people's definitions of AGI — roughly, AI systems that match human performance at most knowledge work (see definition in footnote).
This means that, while the compa
I think there's room for divergence here (i.e., I can imagine longtermists who only focus on the human race) but generally, I expect that longtermism aligns with "the flourishing of moral agents in general, rather than just future generations of people." My belief largely draws from one of Michael Aird's posts.
This is because many longtermists are worried about existential risk (x-risk), which specifically refers to the curtailing of humanity's potential. This includes both our values—which could lead to wanting to protect alien life, if we consider them moral patients and so factor them into our moral calculations—and potential super-/non-human descendants.
However, I'm less certain that longtermists worried about x-risk would be happy to let AI 'take over' and for humans to go extinct. That seems to get into more transhumanist territory. C.f. disagreement over Max Tegmark's various AI aftermath scenarios, which runs the spectrum of human/AIcoexistence.
I consider helping all Earth's creatures, extending our compassion, and dissolving inequity as part of fulfilling our potential.
I don't think that because the aliens seemed to enjoy life much more, and had higher levels of more sustained happiness, that would necessarily mean their continued existence should be prioritized over our's. I wouldn't consider one person's life more valuable than another person's life just because that person experienced substantially more enjoyment and happiness. Also, I am not sure how to compare happiness and/or enjoyment between two different people. If a person had 20 years of unhappiness then suddenly became happy, maybe their new happiness (perhaps by putting all the previous years of their life in a more positive perspective) makes up for all the past unhappiness they had.
If the aliens never had wars, or hadn't had one for the last two thousand years, it would seem incomprehensible to favor our own continued existence over their's. If there were only two possibilities, our continued existence or their's, and we favored our own existence, I imagine that our future generations would view our generation as having gone through a moral catastrophe. Favoring our own species would have robbed the universe of great potential flourishing and peace.
A justification for favoring our own species might be that we expect we will catch up to them and eventually be even more happy and peaceful than they are, and/or live longer in such a state than they would. We would have to expect that we would be more happy and peaceful, and/or live longer in such a state, and not just equally happy and peaceful, since the time spent catching up would add harm to the universe and make the universe overall less better.