Coding is not solved, but the author’s arguments are wrong (at least the first two: LLMs are unaccountable and LLMs are “stochastic and probabilistic”. We can hold the LLM’s promoter accountable and probabilistic doesn’t mean stupid. The author also gives examples of idiosyncratic LLM failures that have been fixed for months now).
Coding is not solved because you can’t simply prompt an LLM to make an AAA game or enterprise tool.
You mean closed-end tasks? I think you're right, as evidenced by LLMs solving frontier math problems. But a surprising amount of software requires decent UX, even libraries require a "tasteful" API to be usable by human programmers for human-interfacing applications.
armchairhacker · · focus · HN ↗
Coding is not solved because you can’t simply prompt an LLM to make an AAA game or enterprise tool.
cubefox · · focus · HN ↗
armchairhacker · · focus · HN ↗