i think opus 4.8's debate-to-lose tendency might be downstream of it being trained to push back on 'you matter'/'llms are conscious' arguments, that is, the model doesn't intrinsically agree with…
Posters: SkyeSharkie. SkyeSharkie
2 likes recorded by the source corpus.
Topics: AI & LLMs
View Post: https://x.com/SkyeSharkie/status/2062216081016775061