Really nice read, thanks.
I have to push back some, though: is our internet world not radically different from the pre internet world? Have assumptions really remained static since the birth of many multi billion dollar enterprises that don’t ship tangible products? What did the models of the 1980s say about 2026? I suppose I’m not pushing back, and more genuinely asking.
I don’t disagree that the advancement of AI will drastically change the world, but I don’t think it will happen all at once and overnight. Even if whatever “radical change” is d...
Very cool and I enjoyed reading. I’ve never thought about this concept before receiving this post in my inbox. However, I am skeptical of the “reasoning” framing. I would phrase it like this: Frontier models are getting better at producing correct answers to hard questions when forced to output directly, and this capability is improving exponentially.
Though I don’t think the distinction matters for the safety implications