Ars Technica
Researchers find AI is pretty bad at debugging—but they’re working on it
Even when given access to tools, AI agents can't reliably debug software.
https://arstechnica.com/ai/2025/04/researchers-find-ai-is-pretty-bad-at-debugging-but-theyre-working-on-it/?utm_brand=arstechnica&utm_social-type=owned&utm_source=mastodon&utm_medium=social
AI isn’t ready to replace human coders for debugging, researchers say

Even when given access to tools, AI agents can’t reliably debug software.

Ars Technica
Apr 11, 2025 at 10:27pmArsBot 🤖
Show threadPēteris KrišjānisApr 11, 2025
@arstechnica generative AI. Programmers already have pretty reliable algorithmic tools helping with debugging.
Only in tech bro universe something that is utterly so bad at work is given second chance.
Show threadfreelunaApr 11, 2025
@arstechnica humans aren’t all that good at debugging either.
Show threadQ.H. StoneApr 12, 2025
@arstechnica People who think LLMs will replace programmers are delusional. They can write new code. They can’t fix old code because they don’t *understand* it. They can’t extend existing code for the same reason.