Michael Stapelberg 🐧🐹😺 on Nostr: Current LLMs have impressive quality (see e.g. for a recent example). I thought this ...

Current LLMs have impressive quality (see e.g. https://mas.to/@zekjur/115775753642030596 for a recent example).

I thought this article was a good read regarding how human and machine intelligence / skills / cleverness relate: https://www.thealgorithmicbridge.com/p/the-shape-of-artificial-intelligence

If you know of other insightful articles or papers about this topic, please share

quoting
note1q5k…ptv9

Out of curiosity, I pointed Claude Code at the problem.

I provided it the backtrace of the crash and a clone of the screen repository.

This was sufficient for Claude to diagnose the issue by itself, provide steps to reproduce (which do indeed trigger the bug) and locate the corresponding upstream bug report.

(Before you ask: the upstream bug report was filed after Claude’s model cut-off date.)

The quality of this tool is transformative, especially in an Open Source environment!