If jury is still out on positivity, long term, of AI, I'd really like to see arguments for that. Historically all - almost? - technical improvements were net positive; even some blunders had upside. AI is dangerous, yes, but e.g. fission was developed for the bomb, and now powers significant numbers of households worldwide - the tech less than 90 years old.

> all - almost? - technical improvements were net positive

I think it’s very likely AI is a technical improvement. But there is still a chance it’s a small improvement being massively overbuilt.