bbmatryoshka

In this game you "win" only if you arrive first, so keeping the same headcount to run faster is a valid option for Anthropic. In this context "winning" means mainly that we lose our jobs

inside their head the vast majority of mathematicians are fuzzy and probabilistic as anyone, the order and rigor is only the tip of the iceberg

If we have a jagged frontier inside the domain of math (very plausible, the jagged frontier concept seems to be quite fractal), the model solved the easier problems in his area of major strenght, some of them will be in the bottom 5% of the easiest problems in the full set, and some in remaining 95% of problems

49% of that money goes to taxes

I (vibe) made an AI lab manager game, I am surprised there aren't many others

non deterministic abstraction are absolutely useful, outside of software sector they have been used since the start of civilization ("a worker" is a very very non deterministic abstraction, outside from the most basic tasks)

also fire is not particularly surprising, once discovered

you can say the same of google, or libraries, information technology is the technology of moving and rearranging information, not of creating it

without Iron Man's plot armor, Ultron would have won

two istances of the same model, with low temperature, are infinitely more similar than any couple of human minds

because human engineers are not reproducible, so having an accurate measure of the skills of one of them has a much lower utility return (those benchmarks are very expensive to run, and making them for humans will be at least as much expensive)

with a good enough LLM you can greatly increase the capability to build a functioning well (and other basic infrastructures) in remote and undeveloped areas without great technical knowledge, they are very good as assistants for practical tasks where you can give feedback step by step on what is happening

fertility is NOT a measure of happiness

2 hours... of agent work, usually for the same output the average worker has to use much more time

Those analyses tend to not consider the government interest in having AI as a weapon: now AI labs are fully part of the military sector, making them much more supported by the system

You are ignoring LLM-ergonomics, some time ago I saw benchmarks showing that popularity the language (and so more data available in training date) was strongly linked with LLM's performances, with top results with javascript and python. I don't know if a year later this is still true, but is absolutely possible

Usually the elephant in a room is something very evident about which no one wants to discuss about