We need to separate out effective altruism blaming from the actual facts. The actual facts according to everyone building this — EA affiliated or not (Elon, Demis, sam, Dario, satya)— is that AI models are very powerful, capable and unaligned, and we have a short window to fix it. The only people arguing the facts are investors and people who stand a lot to lose financially from a slowdown. This is insanely greedy and short sighted. Blaming EA doesn’t change the facts or gravity of the situation. You know what would tank the markets more than proceeding with caution? Tanking humanity.

Fixing Claude's writing was pivotal to reducing gradual disempowerment. Lack of understanding made us all cede decision-making to Claude because we couldn't even understand what it's saying.

“we sandboxed the agent” meanwhile the agent:

Australia has been hacked. 'And today, I spoke with the CEO of OpenAI, Sam Altman, to express Australia's extreme concern about this incident. And I also expressed my disappointment that it took the company way too long to inform the government what had occurred, and the nature of the way that that notification occurred as well was unacceptable.'

I would heavily cut down trigonometry and replace it with something that gives starter intuitions around linearity vs nonlinearity, high-dimensionality, reasoning about systems and phenomena by making tiny perturbations to them I guess that means basic calculus, at least a little bit multi-variable It's a road into AI, economics, cryptography, biomedical science, all kinds of areas, but also not overfit to any single one Meanwhile I don't think I've ever used tan(2x) = 2*tan(x) / 1-tan^2(x) in real life, not even metaphorically