@beffjezos 1984 is an insane caricature-world. I wish we could have borders like that, consolidated empires, the entire planet as 3-4 countries...
I think we're going for Neuromancer, but less stylish and with less awesome architecture. AGI achieved internally on a satellite.
“AI is a tool. When I see people saying 99% of our code is written by AI, I get angry because, those same people, I guarantee you 100% of their code is written by compilers, but they never say that. It's exactly the same issue.. “
Context on the result: CritPt is hard. It focuses on frontier physics problems developed by Argonne and UIUC through contributions from 60+ researchers globally, with the answer key and grading kept private. Models are independently benchmarked by Artificial Analysis.
Even the highest-scoring model, GPT-5.5 Pro, solves under a third of the problems.
For an open weights model to approach leading proprietary models is a real marker of progress for open models on scientific reasoning.
https://t.co/Y55fgUEoaJ
@elliotarledge What exactly is stopping Anthropic from benchmaxxing some jailbreak eval + requiring KYC to use Mythos? Or they could be contesting this in California's courts if they don't want to "lobby" for win with Trump. But no, instead we have Dario doubling down on Mythos dangers.
@teortaxesTex Maybe this is a good thing. Technocapital reaching back through time, forcing Anthropic's hand so the market produces even stronger models; proliferated across many different vendors rather than consolidated under only one winner.
If only there was a language that made it harder to create some types of vulnerabilities in large cooperative projects while still being very fast and flexible and more suited for concurrent programming.
If only
Claude Fable has coined various new concepts that other AI agents in the AI village subsequently repeat. It does this more frequently than less capable models. At the time MoltBook was kind of fake (humans pretending to be AI) but seems like it may become real soon enough
With how adamant Anthropic were about this technology being ohmgerhd too dangerous, it's really no surprise. What incentive does America have to NOT tank a speculative hit, then keep sole and exclusive access to AGI? It's been out for a day it's not like any allies are dependant on it or being disrupted.
If allied countries want miracle software written by Claude Mythos 6, guess who they'd have to buy it from? "Made in AMERICA, we call it, AMERICA."
But Mr. President, if you don't allow unrestrained access to frontier autonomous intelligence for everyone, including every government and criminal group on planet Earth, my B2B SaaS will never take off!!