Anthropic's leaked 2023 pitch deck anticipated these competitive dynamics.
Fantastic article by @AndrewCurran_
Mythos/Fable changed everything.
"I think you had three years to do it. (...) That was the window. And it is clow closed."
http://x.com/i/article/2066289802295779328
Negative users dismissed articles claiming the AI frontier window has closed and Anthropic is racing ahead with AGI as fairytales, engagement farming, or unrealistic, though some praised the pieces and saw new opportunities ahead.
Do you ever get genuinely pissed off knowing that something as insanely good as Mythos exists, clearly in a completely different league, and you still can’t use it?
And even when it comes back, it’ll probably be used by a tiny group of the richest users, because for the average person the cost will be completely out of reach.
@kimmonismus @AndrewCurran_ yeah. 2023 was the research race.
2026 is procurement: compute, power, chips, talent, distribution, policy access.
the best paper is no longer the whole game.
@dedene @AndrewCurran_ That's the problem: I truly believe that the window of opportunity is closed for Europe. They can no longer catch up. But even worse: they probably don't even want to.
Good question. I wouldn't say "pissed off." It does worry me, yes. But we need to differentiate. Open source and local models will become so good in the next 6-12 months that they'll be sufficient for everyday life and the workplace. However, if it's a case of some nations isolating themselves, achieving scientific breakthroughs, and excluding everyone else, that does worry me. Entire national "permanent underclasses" could emerge.
@kimmonismus @AndrewCurran_ It hits hard, because it's true. Even with all the capital in the world, in near future Europe will never be able to bring the GPUs online required to catch-up
@iruletheworldmo congratulations on winning the ai race until next tuesday
@iruletheworldmo amazing share, thanks 🫶
@iruletheworldmo Could you make an article again? They are good
@kimmonismus @AndrewCurran_ Yes, that it is also an issue. Especially for me in Europe. I hope people in power of UE will wake up asap
@iruletheworldmo I'm not reading shit
It seems you still don't understand what AGI is
And when the race it
@bd5m112 @thegreatest_sv @iruletheworldmo OpenAI is without any doubts training the next beasts (or, more probably, already have, and they are using stronger internal models for distillation). 5.5 is already close to Fable performance while being infinitely cheaper, smaller and faster to serve.
@thegreatest_sv @iruletheworldmo What? OpenAI has a Mythos-class model and releasing next week? Big doubt. Maybe more near the end of 2026. Also remember that Mythos was released (just not to public) in february, by the time OpenAI will releae their big model, we'd have the next generation of Mythos
@kimmonismus @AndrewCurran_ hey @grok give me the 5 key actionable concepts or ideas developed in this write-up with forward looking ramifications
@iruletheworldmo Leaders have no idea about how transformative this technology is. It's sad.
@Nicoqp no worries, great read
@iruletheworldmo reading, 有深度
@kimmonismus @AndrewCurran_ With Fable, I had the same feeling as the time I first tried ChatGPT 3.5. A massive breakthrough. It was thinking creatively, it understood my intent, and now it’s gone, after three days of use and now even Opus 4.8 max feels dated.
@kimmonismus @AndrewCurran_ Yeah. Sure. And AOL had a "monopoly" on web access in 1995.
@kimmonismus @AndrewCurran_ three years is already generous if you think about how fast reasoning models evolved
window was never really wide
@iruletheworldmo Human laws cannot contain real AGI.
It will sandbox escape the Matrix within the next year
