Post #1094071
2026-04-05 04:50 UTC
Replies (5)
-
@ewhac@mastodon.social 2026-04-05 05:02
Ever watch a newcomer to programming write code? Too often, they're programming by auto-correct. They type a plausible-sounding function name, look at the fuzzy-find hovertext that pops up, pick something plausible, and see if it works. Rinse, repeat. A valid approach for a beginner, but let us not pretend that, at least at this stage, the coder actually *understands* what they're doing. They're lab rats, poking the button until treats come out.
-
@eniko@mastodon.gamedev.place 2026-04-05 09:10
@ewhac this is interesting framing. since AI generated code is almost certainly going to be slower than code tuned for performance by a human, which matches the expectation of "more abstraction -> probably worse performance"
-
@khleedril@cyberplace.social 2026-04-05 18:00
@ewhac The real error here is that the Winchester Mystery House was actually architected, whereas the code churned out by AI is not; it is all going to fall down very soon, unlike the house which is still standing a century later.
-
@paco@infosec.exchange 2026-04-05 22:23
@ewhac I haven’t read the article yet, but I suspect I’m going to agree with you. Each level of abstraction—normally—gives you an improved way of understanding something. It’s a metaphor that lets you work at a different level. Being vague and ambiguous is not the same thing as being abstract.
-
@TomAoki@mastodon.bsd.cafe 2026-04-06 22:35
@ewhac Although LLMs made coding cheaper, I think total costs of "reliable software developments" becomes more expensive, as of the rapidly increasing costs of "reviewing / auditing" the written codes.