It's crazy, you see someone spin up 50 instances of chatGPT or gemini or whatever to tackle a math problem, it succeeds by brute forcing through a ton of existing theories to find one that extends the problem, and the takeaway is that this technology is magic and going to solve all of our problems.

Whereas I see that and say - if we properly funded the sciences we could have had a bunch of grad students tackling that problem and found this application 20-30 years ago. Sure it's 'nice' that LLMs can fill in for people in brute force work like that but people are perfectly capable of doing that work and if we focused on properly staffing our research institutions we would achieve a lot more a lot faster. Instead this is obviously going to be used to replace staff and further reduce headcounts.