D and AI
monkyyy
crazymonkyyy at gmail.com
Mon Feb 16 16:21:46 UTC 2026
On Monday, 16 February 2026 at 16:14:52 UTC, jmh530 wrote:
>
> If the problem is complicated enough, then ChatGPT5.2 thinking
> still barfs on it. What seems to help is when the AI can
> actually run DMD.
Thats true of all ai's, if the chess ai's hallucinate 20 possible
moves, probably at least 5 of them is a blunder.
Without filtering probabilistic computation will always be
worthless.
More information about the Digitalmars-d
mailing list