Obviously the right thing to do. How to reliably do it might be a bit difficult, but fortunately a ban doesn't have to be 100% effective to be beneficial.
There is going to be something that would not have existed without LLM because there wouldn't have been enough resources to create it before. Probably already is. Some of it will be amazing, although most of it probably not.
Example: Qobuz has no official Linux client, but there's https://qbz.lol/ which it seems to me is mostly vibecoded. It's not perfect, but it works pretty well and can do some fancy stuff like bit perfect DAC passthrough up to 192kHz/24bit.
I'm on the same page with you that their AI stance is probably harmful for them -- but I think they wrote those as a reaction to the contributions from Bun and others. So your implied timeline is not entirely correct.
I think the mistake people are making currently is that they publicize stuff way too early. I.e. they make a prototype and then put that out there. Possibly with a full product page and all.
People did this also before LLM, but the difference is that now the prototype is a fully functional product in the technical sense whereas before it was little more than a glorified Hello World. The human effort and calendar time used remains roughly the same.
There's gonna be amazing products made with LLM, but it'll take some time. Not as long as it did before, but still significant time.
Please oh please try to make Kärcher adopt your stuff. Even their latest premium cleaning bots are hopeless when they don't know where they are, even when I tell them where they are.
I read the article for 5 minutes and spotted several instances of AI slop in the writing style. It's not that difficult to make your agent write like it was a human.
Yas. Only thing I've been able to practice in any real way. I've switched disciplines often though, so haven't got very far in any of them. I got a belt grade from like 7 pretty different japanese, thai and korean styles, but my highest grade is judo 3. kyu from like 1996.