[k-thimmaraju]

I ran the data through our LLM behavioral analysis system at splabs.io, and it alerts Red on the RLHF-optimized output compared to a Yellow on the no-RLHF.

You can check out the analysis here: https://splabs.io/ai-psychosis-and-cognitive-cost