My problem with all these fantastical predictions, is that most of these real big problems aren't really technological problems; they're social and political problems.

In order for "AI to solve X", there has to be a tremendous amount of resources put into not only AI solving X but testing and implementing the AI-generated solution(s). Who is going to put in those resources? The money flows towards what's profit-generating, not problem-solving.

So far, AI hasn't solved any problems that matter in terms of improving the human condition in any significant way. What has it accomplished so far? It has allowed companies to develop software faster so they can cut their workforce. It has solved some tricky math proofs. That's about it. Neither of these make a dent in improving humanity's lot.

"Yeah now it's just a chatbot, but once it gets better, it'll solve X." Yeah, sure.

_Theoretically_ AI could develop solutions to some gnarly technological problems -- nuclear fusion at scale would probably be at the top of my wish list -- but as we've seen the last 2 years in the US, the main reason we aren't relying more on cheaper/renewable energy sources is political, not technological.

Medicine is similar - the reason we still have Ebola outbreaks in the DRC isn't because we don't have a cure for Ebola. Same for the half a million children who die from malaria each year; it's not like we don't know how to stop malaria. (If they were American children, we would have have eradicated malaria by now, but they're African children ...)

World hunger: we know how to produce enough resources to feed and clothe the planet -- it's just very badly distributed, by design.

Etc.