RE: https://toot.cat/@zkat/115860108367572702
I am not dissing this idea or list at all. I like the idea of it, and all the use cases currently on display for uses of AI /LLMs are some of the worst examples of using it, making LLMs totally shit at many things. I definitely do not like the “Jesus (in the form of LLM ) take the wheel” idea of coding. It lessens everyone by robbing us of learning opportunities and the joy of discovery and mastery of a skillset.
That said , I think there are reasonable practices and use cases of true AI Assisted coding , stress on “Assisted”, where the ultimate last hands on the code should be a human be it in the form of a thorough review and testing. This won’t reduce the amount of bugs that slip through but how would that be any different than now?
This comes from looking at the KeePassXC response to concerns of them using AI/LLMs, and their response seems pretty reasonable to me, where nothing that goes in, is not reviewed and they’re very open about its usage. Whether or not this actually plays out in practice, time will tell.