yes it’s good for acceleration. It’s not a safety org. trump would never brag about “safest” on anything.
If they really form an AI force, they can just nationalise Anthropic and OpenAI. Then data centres and constructions will be under federal jurisdiction. All of these protests wouldn’t matter
Day 19 of occupying OpenAI's NYC offices until OpenAI pauses frontier model training
I have updated the demand in light of the severity of the situation. OpenAI (and all AI companies) needs to unilaterally pause frontier model training immediately. Any new model could be the one that kills us
- Met and had good conversations with a few more employees
- Changed a conservative guy's mind about existential risk (he appreciated the framing of "we order China to stop too, and if they don't, we bomb them")
- Tried out livestreaming for the first time. It makes people less likely to want to talk, so a bit of a mixed bag
Good day
@Actuallykeltan Have you noticed something? All the escapes and “rogue” hacking incidents are from within the labs themselves? They are converting their failures to “loss of control”
@Actuallykeltan He literally said that Western labs should advance so far that China would have no choice but to negotiate. He always mentions CCP in particular. They'll never agree and believe this slowdown nonsense