> I wouldn't agree that LLMs will help write better software

Your statement makes no sense.

Even if you don't let LLMs author a single line of your code, they can still review it, find edge cases you didn't think about or suggest different approaches.

The fact that AI allows lots of slop, does not negate its overall utility in good informed hands.

▲

sesm 2 hours ago | parent [-]

The balance between 'find edge cases' and 'hallucinate non-existing cases and waste your time' may be negative. LLMs are also not free, they cost significant money even today, when they are subsidized by marketing budget.

▲

epolanski 2 hours ago | parent [-]

Even an hallucinated non-existing case (which isn't the norm in my experience) is still food for thought.

▲

Madmallard an hour ago | parent [-]

It was already shown repeatedly in GitHub repositories in the last year that authors are really unhappy with AI generated pull-requests and test cases.

	▲	epolanski an hour ago \| parent [-]
		I am not invested in anything, I am merely sharing my personal experience.