Most people cannot handle facts and truth. They want comforting lies instead. Humans are crap like that.

That just would mean even more tokens wasted for worse results.

Vibe coding is to actual building of software as putting something in the microwave is to actual cooking. Only that the microwave results are orders of magnitude better.

Face it, coding is hard. Have the talent, invest the time or you will never be good at it and no tool can fix that.

"Incel" is not referring to a man without a woman. It refers to a specific toxic community in that state.

There are some left. And those are special enough that AI cannot do the job.

That is also what I hear from people that have tried it. For example, funding applications for research: 10% hard content, 90% crap. The 90% can be supplied by AI because they are never read anyways, just checked that they are there. And LLMs can do "better crap" nicely. But for the 10%, no chance to have an LLM do it successfully at all.

Many humans are good at hubris. The dumber and the less skill to fact-check, the more hubris.

Simple: All of those that can reasonably be automatized, have been so already, because AI is neither needed nor a good idea. The ones left are the ones where automating it does not work for some reason. This can be because they are special, or need a lot of maintenance or AI hallucinating even once would be a major disaster and conventional automation does not cut it.

Yep. Humans are bad at routine, boring stuff, but far, far superior to automation at anything unexpected. And that is not going to change anytime soon.

Safety reasons. Same reason you have two pilots in any larger plane. There is just too much stuff that cannot be predicted or is too rare to make ecconomic sense to automatize.

Indeed. Whoever wrote that has no clue at all. Passenger Attendants are there for safety reasons and they cannot be reduced without endangering lives. The "service" aspect is merely a side-benefit.

And then the good translators come back. That will happen for a lot of the specialties that "AI" is supposed to replace.

I can see all those "business-something teachers" vanishing though. I did ask ChatGPT for some slides for my IT Security lectures and all my students and me would get out of those is a good laugh. Had an eerie resemblance to some business teaching slides I have seen though. Facts, no explanations, no systematic, no support for understanding, just raw knowledge to remember. And too simplistic and sometimes outright wrong. Useless for STEM.

It really does.

Indeed. Good historians are these people that tell society not to make the same fucking stupid mistakes time and again. Say, like putting authoritarians or criminals in power.

Mathematicians really understand what some math tool does and does not do and hence can apply it competently. They are basically the only people that can do that and AI/LLMs are not even in the run for that. In limited areas theoretical non-math STEM people can compete, but they are typically even more expensive and hard to get.