4 ms·

Perhaps, though I'm not convinced that incentives really work like that. For every Elon Musk who makes grandiose promises and then either fails to deliver at a

Perhaps, though I'm not convinced that incentives really work like that. For every Elon Musk who makes grandiose promises and then either fails to deliver at all, or eventually delivers 9 years after he promised, there must be multiple senior execs who would be sufficiently embarrassed to make a bold prediction about the near future that utterly fails to materialize. Certainly these people have already seen the capabilities of what will be released in the near future after all the safety boxes have been checked. Certainly they know more about the likely near future than me or 99.999% of everyone else. So when I hear a bold prediction about the near future that's essentially a synthesis of opinions of this handful of insiders who are in the best possible position to know, why would I discount it by 100% simply due to hand-wavy "incentives"?

ein0p 2y ago
I _know_ what will be released in the next year, year and a half. It’s basically just improvements to models, multimodality, multilinguality and various optimizations. Absolutely nobody at my current FANG employer is currently working on anything even remotely resembling “AGI” or indeed anything qualitatively different from the technologies we already have.
- panarky 2y ago
  > It's basically just improvements to models ... Even though I'm expecting to be shocked by what use cases those improvements will unlock, I expect it will be even more shocking than I expect.
  - ein0p 2y ago
    Not really, no. Much of the slack has already been taken up in GPT 3.5 era, and things haven’t dramatically improved since GPT4. 4o basically alludes to what I said above: it’s smaller and faster than 4, but the level of capability is very similar, the next big thing in it is multimodality, which is relatively straightforward to add. It just makes sense to pay attention to cost and efficiency now. It’s incredibly expensive today to both do research in this field and deploy the results of that research. Even FANGs are severely compute constrained to the point where when Google was training the bigger variants of Gemini I couldn’t even get a handful of TPUs scheduled to do my work (I’m not at Google now). That, quite obviously, is untenable. But even if NVIDIA/TSMC somehow tripled their production and everyone got as much compute as they can use, there’s no serious research going on for how to do cognition. The closest we got to that is neural network assisted Monte Carlo a-la AlphaZero and the various other approaches to search. But all of them are very narrow. Nobody has even a foggiest clue on how to bring anything cognition-like to the general domain. And don’t get me wrong, I’m a bona fide AI enthusiast, I’ve been in this field for over a decade. It’s just that all these promises of singularity “real soon now” are nothing more than clickbait.
rsynnott 2y ago
> For every Elon Musk who makes grandiose promises and then either fails to deliver at all, or eventually delivers 9 years after he promised, there must be multiple senior execs who would be sufficiently embarrassed to make a bold prediction about the near future that utterly fails to materialize. I mean, to be clear, Schmidt hasn’t been a senior exec in the normal sense for 13 years; that’s when he stood down as Google CEO. But also, CEOs predict complete nonsense _all the time_. That’s why you’re not browsing Google+ through your VR headset while heading home from your job at the vertical farm in your self-driving car, for a night of watching 3D TV and drinking Juicero.

ein0p 2y ago

I _know_ what will be released in the next year, year and a half. It’s basically just improvements to models, multimodality, multilinguality and various optimizations. Absolutely nobody at my current FANG employer is currently working on anything even remotely resembling “AGI” or indeed anything qualitatively different from the technologies we already have.

panarky 2y ago
> It's basically just improvements to models ... Even though I'm expecting to be shocked by what use cases those improvements will unlock, I expect it will be even more shocking than I expect.
- ein0p 2y ago
  Not really, no. Much of the slack has already been taken up in GPT 3.5 era, and things haven’t dramatically improved since GPT4. 4o basically alludes to what I said above: it’s smaller and faster than 4, but the level of capability is very similar, the next big thing in it is multimodality, which is relatively straightforward to add. It just makes sense to pay attention to cost and efficiency now. It’s incredibly expensive today to both do research in this field and deploy the results of that research. Even FANGs are severely compute constrained to the point where when Google was training the bigger variants of Gemini I couldn’t even get a handful of TPUs scheduled to do my work (I’m not at Google now). That, quite obviously, is untenable. But even if NVIDIA/TSMC somehow tripled their production and everyone got as much compute as they can use, there’s no serious research going on for how to do cognition. The closest we got to that is neural network assisted Monte Carlo a-la AlphaZero and the various other approaches to search. But all of them are very narrow. Nobody has even a foggiest clue on how to bring anything cognition-like to the general domain. And don’t get me wrong, I’m a bona fide AI enthusiast, I’ve been in this field for over a decade. It’s just that all these promises of singularity “real soon now” are nothing more than clickbait.

rsynnott 2y ago

> For every Elon Musk who makes grandiose promises and then either fails to deliver at all, or eventually delivers 9 years after he promised, there must be multiple senior execs who would be sufficiently embarrassed to make a bold prediction about the near future that utterly fails to materialize. I mean, to be clear, Schmidt hasn’t been a senior exec in the normal sense for 13 years; that’s when he stood down as Google CEO. But also, CEOs predict complete nonsense _all the time_. That’s why you’re not browsing Google+ through your VR headset while heading home from your job at the vertical farm in your self-driving car, for a night of watching 3D TV and drinking Juicero.