zlacker

[return to "LLMs cannot find reasoning errors, but can correct them"]
1. ggm+Dg1[view] [source] 2023-11-21 02:24:42
>>koie+(OP)
How does an LLM question it's priors and axioms?

People have to do this all the time. Bringing skepticism to that table "excel made for you" is a vital part of heading off bad reasoning. For an LLM its a given.

◧◩
2. trasht+Zi2[view] [source] 2023-11-21 11:16:56
>>ggm+Dg1
> People have to do this all the time.

If most people you meet actually question their axioms at any frequency, unless forced to by cognitive dissonance, I envy you.

In my experience, people will go to extraordinary lengths to NOT have to question their own most cherished axioms.

[go to top]