Context on prior (as of Dec. 15th) from VerificationGPT:
https://bsky.app/profile/chrislengerich.bsky.social/post/3kgkkfjbwe32u
@ArmandodiMatteo
Candidate 0: Resolved as yes if a predictive model (like GPT-X) trained to accurately predict a stream of consciousness phenomena (like text) achieves at least 60% token-level accuracy predicting a 'omics dataset with minimal adaptation dataset allowed (largely consisting of an text explanation of a shared latent process).
Candidate 1: If we compare a model trained on 'omics data to a model trained on 'omics data + text, we see at least X% improvement.
Candidate 2: 60% of experts agree that regulatory mechanics show signs of consciousness, such as top-level processes being able to suppress lower-level expression, frames and context stacks, episodic memory, etc. This demonstrates a shared latent process.
Do you prefer any of these definitions, or another?