←back to thread

The "confident idiot" problem: Why AI needs hard rules, not vibe checks

(steerlabs.substack.com)
323 points steerlabs | 2 comments | 04 Dec 25 20:48 UTC | HN request time: 0.018s | source
Show context
tasuki ◴[08 Dec 25 15:19 UTC] No.46193203[source]▶
I wish we didn't use LLMs to create test code. Tests should be the only thing written by a human. Let the AI handle the implementation so they pass!
replies(1): >>46193583 #
1. lxgr ◴[08 Dec 25 15:43 UTC] No.46193583[source]▶
Humans writing tests can only help against some subset of all problems that can happen with incompetent or misaligned LLMs. For example, they can game human-written and LLM-written tests just the same.
replies(1): >>46210175 #
2. tasuki ◴[09 Dec 25 20:27 UTC] No.46210175[source]▶
Not property-based tests. Either way, the human is there to tell the machine what to do: tests are one way of expressing that.