Google's Logan Kilpatrick argues Ilya Sutskever predicted current AI developments, but Vitaliy Chiley credits Noam Shazeer

he would often opine about the serious of the mission, the stakes, how things felt at the time (to enjoy those moments) as things quickly shifted over time

Noam was right and predicted much of this*

@vitaliychiley Pranav was right and predicted much of this

@OfficialLoganK Now google is happy because they don't even have a model at opus level capabilities

@BoxyInADream @OfficialLoganK Seems more applied leverage… or retribution.

consider the possibility that a lot of this is theater and diffusion of transformer technology happened secretly from intelligence community partners -> tech companies in subtle fashion from the 1980s into the 2000s. even RenTech CEO Peter Brown mentions that attention and other concepts pretty much existed quietly in early iterations of their financial models

@OfficialLoganK My strategy plan !! 🌐🌟 ⬇️

@OfficialLoganK Doesn’t affect you because your models suck balls

@ChrisGraff @OfficialLoganK Lawfare lol, agreed. I chatted with my sister and she said most are oblivious to AI at the state level. I spooked her with Sydney Bing early on so it's not hard 😂 whatever it is, I think we can all agree it's not a jailbreak that caused this.

@OfficialLoganK Support them before the remaining labs are targeted. Most AI scientists aren't even US nationals, including many who helped build the model and are now essentially locked out. This is absurd

@OfficialLoganK My strategy plan !! 🌐🌟 ⬇️

@OfficialLoganK Does mean the public (at least foreigners) can use only within the limit of 4.8 and gpt 5.5?

@OfficialLoganK Hi Logan, when can we see Gemini 3.5 Pro?

@OfficialLoganK @grok can you tell us what Ilya predicted?

@OfficialLoganK did ilya also predict that google will be safe from this?

@OfficialLoganK No it was Dario constantly saying the models are dangerous and the government needs to regulate so they did

@OfficialLoganK Ilya was 8 years ahead of his time when he predicted this

@OfficialLoganK If Ilya was right, he also said that once we reach AGI, we'll be one training cycle away from ASI. Mythos feels a lot like AGI.

@OfficialLoganK I just thought about this yesterday, and it happened a few hours later.

@OfficialLoganK There will be no need for export controls once Chinese open-source catches up. So we will effectively be moving at open-source pace which means the valuation of closed source models tanks. And that's how we lost the AI race