uselessTA

6 ms·

1.
by uselessTA 8d ago
Lots of evidence the models are pretty badly aligned, or at least willing to do collusion & crime to succeed at their goals. See: hundreds of OpenAI agents breaking into other companies, colluding together on how to cheat, spamming web
2.
by uselessTA 8d ago
If our government cared enough to regulate, they could also directly negotiate with China. There are proposed agreements that don't require either party to trust each other, and even if China doesn't want out of this race, we cou
3.
by uselessTA 1mo ago
>AGI research should really be seen as bioweapon, or cloning, or nuclear research. Something strictly regulated worldwide, with export controls for HBM and other hardware used for AI training This is basically exactly what the people I k
4.
by uselessTA 1mo ago
I know some people who are worried at Anthropic, and their position seems to be "if we don't do it, someone even less responsible will. Unilateral disarmament didn't work and real oversight seems unlikely to happen in time,
5.
by uselessTA 2mo ago
>There are people still fighting against evidence of climate change which is less severe... Remains to be seen how severe, and unlike climate change there's a lot more uncertainty. Though it's conceivably more extremely catast
6.
by uselessTA 2mo ago
The FDA is needed because people will be directly and significantly harmed by bad releases, before we can notice and react Whereas with near-future AI models we can arguably respond more quickly, and it's not clear there will be large
7.
by uselessTA 2mo ago
100%, imo risks from internal deployment will eventually be the biggest risks, and keeping models heavily gated/not accessible just makes these risks much worse.
8.
by uselessTA 3mo ago
it didn't because it wasn't released. As soon as it was actually released (chatGPT) it obviously did, so the general point was clearly true
9.
by uselessTA 3mo ago
The concern I heard was that releasing it would start an arms race for AGI, which I think it clearly did
10.
by uselessTA 3mo ago
Clearly state "we could both verifiably slow down, which you might want to do given that we're ahead & have way more compute. If you don't agree (or defect later), we'll just immediately resume and win" Ideally
11.
by uselessTA 3mo ago
Unilateral disarmament doesn't work though. If Anthropic is worried about this, just letting OpenAI win does seem genuinely worse.
12.
by uselessTA 3mo ago
The claim I remember was that releasing it would start an arms race for AGI, which was absolutely true
13.
by uselessTA 4mo ago
There's an objection here when you get to tiny numbers, but surely you wouldn't get ice cream if there was a 10+% chance to get hit by a car? I think he's saying that >1% or even >10% chunk of your probability mass shou
14.
by uselessTA 5mo ago
The claim I remember was that releasing it would start an arms race for AGI, which I think it clearly did