For ARC-like tasks, o3 is more expensive than a knowledge worker. But o3 is probably still faster.

So if the o3 might be a good option for activities where you want to be at the speed/smartness frontier. Maybe trading?

Obviously, it still depends on your reliability caveat.

It's a useful slide. He is explaining the analogy that motivated work in what has demonstrated itself to be a successful approach.

Hinton says the same thing. Analogies between brains and neural networks inspired his faith in deep learning, when empirical results were still disappointing.

Reading this, he doesn't really explain his own initial psychology very well. Why did he commit an armed robbery and murder? He attributes it to his trauma, but what's the linkage? He asserts that he has made progress toward rehabilitation, but reading his essay, I am not so sure.

Agreed that debt is a stock, but it's not a good measure of the scale of an organization. A big country could have zero debt. Seems better to compare the federal budget to a corporation's expenses and those are known quantities. But that comparison, while more informative, yields a less spicy result

I don't think it makes sense to compare the federal budget (a flow) to the market capitalization of a corporation (a stock). A better comparison would be between the federal budget and a company's expenses, which would reveal the federal government to be a much larger operation.

The "write code with synthetic data" approach sounds great as a new tool, especially if it speeds up research and makes new info available.

But if it replaces more complete access, that seems bad. It's way way harder to do good research if you can't work with the raw data.

By what process would you discover that an existing generic works for a new condition? Would you have to run new clinical trials?

This was great. I am confused how the non-profit can currently control OpenAI, but have so little resources (~$20 million, no staff) with which to advocate for themselves. Why can't they use their control to shift resources in their direction and hire a staff?

If it were real, plant consciousness and sentience would be a painful discovery for vegetarians. And it would ease the guilt of meatlovers. It would be an emotionally high stakes question either way.

But plant sentience seems much less likely to me than plant consciousness.