ai is 70% hype, the remainder is real.
it can code fairly well, you still need to plan/arch/decide and review. devs will just do more work, instead of six projects they will have 12. they will work you more than ever.
it can write nicely, thats fine. Vision works, audio/tts works. transcription mostly works. but that's so small that productivity grows only fractionally.
anything deterministic is a sh-t show, hard to build, hard to maintain. complicated. solving business problems is even harder as most folks are trying to use ai to fix problems that cannot be fixed, messy data, erps fragmented, bus processes broken. this is why you do not see big wins on the enterprise side (aside from things that vision, automation, etc. can do).
so yeah, there is value but it's far from 'ai replaces everyone'. it replaces certain things that would have been replaced with automation anyway + coding.
so yeah, it's 70% hype. it'll calm down after venture folks and tech bros get their $$$ back - at that point we are going to a traditional model where things happen gradually and over time, just like things that happened with the internet bo-m, cloud, mobile, etc.