(steerlabs.substack.com)

323 points steerlabs | 1 comments | 04 Dec 25 20:48 UTC | HN request time: 0.22s | source

Show context

jqpabc123 ◴[04 Dec 25 21:38 UTC] No.46153440[source]▶

We are trying to fix probability with more probability. That is a losing game.

Thanks for pointing out the elephant in the room with LLMs.

The basic design is non-deterministic. Trying to extract "facts" or "truth" or "accuracy" is an exercise in futility.

1. jkubicek ◴[08 Dec 25 16:46 UTC] No.46194503[source]▶

The author's solution feels like adding even more probability to their solution.

> The next time the agent runs, that rule is injected into its context.

Which the agent may or may not choose to ignore.

Any LLM rule must be embedded in an API. Anything else is just asking for bugs or security holes.

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