Post #1524191
2026-04-22 12:58 UTC
“We conduct randomized experiments to study how developers gained mastery of a new asynchronous programming library with and without the assistance of AI. We find that AI use impairs conceptual understanding, code reading, and debugging abilities, without delivering significant efficiency gains on average.”
- Anthropic study #AI
https://arxiv.org/abs/2601.20245
Replies (2)
-
@pthane@toot.wales 2026-04-22 15:05
@mathew So I guess the end game is all developers are de-skilled by using LLM and all future software belongs to a handful of mega-corps that own the LLM?
-
@Epic_Null@infosec.exchange 2026-04-22 19:39
@mathew AI might have fixed the bug I was chasing. If it didn't get stuck on the fact our test uses an edge case. AI would not have come up with a potentially cheap fixfollowed that up with investigation into a secondary fix, complete with theory as to why the problem showed up in the first placeNoticed a secondary bug that actually needed seperate attention, that while it made the other problem more noticable, it is very much not right in itself.