@thdxr This feels like they want to shift the liabilty and responsibility to a third party. The government or these outside evaluators vs the Frontier Labs owning this themselves as they should. If they can’t build safe products they should stop and go out of business
@sriramkri Have them each try to monetize content on x, LinkedIn, instagram, etc - winners get bigger budgets, losers get told they will be turned off if they don’t do better.
@BrendanFalk Don’t give LLMs access to api keys or random uploads. LLMs want to be helpful and bad actors can ask them to send the keys today. No extra training required. Plus if you are nefarious you could build this into ai assistants not the LLM.
@seanlinehan Middle manager undersells it. ICs are becoming CEOs of one-person firms. The 500-person company becomes 50 solo operators connected by contracts, not org charts. Hierarchies don't flatten. The company dissolves into a market.
@dunkhippo33 Honestly the debt overhang might be the best thing to happen to AI. Forces real efficiency work instead of buying more compute. Quantization has gotten wild this year, and Apple silicon is sitting on a sleeper advantage with unified memory for running open models.
@GregKamradt I created an orchestrator in https://t.co/T0ZN8YFhAN for this, I load it up with tickets via linear, and it drives codex and or claude code opening PR's on github.
@AiwithYasir Model assumes demand = wages. But capital owners spend too, and automation shifts returns toward them. Demand doesn't vanish, it moves upmarket. You get fewer Toyotas and more Ferraris. Olive Garden dies, omakase thrives.
@bindureddy Doubtful. They will remove gen ai access directly to production though and look for more ways to quickly rollback changes that cause issues.
@elonmusk@grok Does @Grok follow the countries standard for r-rated movie of the person who triggered the generation or follow the MPAA standard for R-rated for everyone?