Is this something everyone is noticing?
Are there some well known folks who are producing things using AI?
> I think in three years AI will become useful for mathematicians. It will be a great co-pilot. You’re trying to prove a theorem, and there’s one step that you think is true, but you can’t quite see how it’s true. And you can say, “AI, can you do this stuff for me?” And it may say, “I think I can prove this.”
> It’s a very exciting potential use of AI to create connections or at least point out possible connections. Right now it has a very lousy success rate. It might give you 10 suggestions of which one is interesting and nine are rubbish. It’s actually almost worse than random. But this could change in the future.
As usual the title is embellished and his actual interview is it "might" be useful.
The hard part of programming—the expertise that comes from experience—happens when gathering and defining requirements. Writing code takes skill but got more automated over time, from compilers to IDEs to LLMs.
Getting to “almost working” with AI coding looks impressive, but like a car missing only one wheel it has zero utility. AI coding looks impressive to people who could not write code at all. It seems less impressive to those who can.
I won’t call it useless but at this point the time spent chasing down hallucinated library functions and API endpoints, and auditing code for security problems and missing edge cases more than offsets any benefit for experienced programmers. Maybe it will get better over time.