No, I'm not. First, you used a science fictional computer to make a point, next, you used a summary from AI that read much like the contents of this thread. It was a fair summary of the rumors floating around. I suspect your expertise is something other than writing software. I don't pretend to be a code expert, but I still need to understand it reasonably well.
These AI systems, such as Chat GPT and the others, aren't flying your plane. The AIs you cited are very different software from almost anything else you use. The software for our planes and commercial jets was created prior to AI becoming popular. People are using AI to write code, but capable people still look through the code. The AI generates exactly what someone asks for, but won't handle exceptional circumstances unless those were part of the query. In some ways, AI creating code may be better as it will "close" code blocks better. I've seen bugs from people failing to close loops or decisions. The compiler or interpreter should have caught them, but there was sufficient ambiguity that the problem appeared at run time. I've noticed bugs when a user entered data that was interpreted as a hex code, which later caused memory leaks.- an abbreviation for 10% ethyl acetate was entered as 10%EA which the system interpreted as something else. I don't know if AI generated code would have handled that any better or worse.
The failures are lapses in training, people cutting costs, or an unanticipated problems. In a small plane, there are some options at 600 feet, but in a 787, the best option with dual engines out is simply to pick the best place to crash because there isn't time for anything else. As we don't know whether the automation failed, this is a straw-person argument.
The 777 and earlier aircraft seem to do most of the flying in the same fashion. This paragraph contradicts your earlier comments as I parse it, especially the last sentence.
