NB: I'm talking about skill cap here, not speed of execution. Of course, an AI will be faster than a programmer… *if* it can handle the job, and *if* you can trust it enough to not need even more time in review…
Your point is valid.
"AI leads to loss of personal coding skills"
Unfortunately, I can no longer do long division. No one will pay me to do long division and I have a calculator now. I could stay sharp at long division for a hobby though. Keep those for loops sharp if you want, but I don't see people paying you to hand code. Eventually, it will just be a liability. (like not using a calculator).
"it could hit a hard wall at 70% of a good programmer's ability"
That is not what NVDA,AMZN,GOOG,or MSFT believe. Maybe you are right and they are all wrong. They do have some smart people on staff. But, betting against the sp50 is generally a terrible plan.
Well, personally speaking, I'm paid to hand code; LLMs have not reached the quality of my code output yet and I'm seeing no pressure at all to use LLMs.
Relatedly, I work on an open source project where the constraining resource is review (as it is in most open source projects.) The current state is that LLM generated code is incredibly hard and annoying to review and there is a lot of pushback.
So, I'm going to wait and see.
(...especially since there's also legal challenges to LLMs trained on open source code with no regard to its licenses.)
Long division is a pretty simple algorithm that you can easily and quickly relearn if needed even your LLM of choice can likely explain that to you given there's plenty of writing about it in books and on the internet.
It is all about money and power
Aka pub and mass control (/propagande)
Another good reason to avoid getting raped by AI (but nothing related to the topic at hand (code))
"Are you using AI ?"
(his response, tldr: "yes but actually no, because it sucks")
Great collaborator