>>33794an llm can't really come to any new conclusions or ideas, it's stuck with everything that was put into it, those patterns, which are mediocre, really. they're basically relying on what's effectively fancy autocompletes to do things for them. if they need something new, or something breaks, or there's some hilariously blatant vulnerable code, they won't be able to do anything because they don't actually understand any of it. they're literally stuck with random numbers.