A structured methodology for capturing how human cognition diverges from model reasoning — preserving both sides' logic, not just the outcome. Where RLHF captures that a human preferred output A over B, the corpus captures why. That reasoning is where the information about human cognition lives. At population scale, individual corpora surface generalizable patterns in model miscalibration.
Status, September 2026: this idea shipped inside Novalyst as its About You corpus — calibration entries, preferences, and lessons that every AI session receives. The methodology is now a product feature rather than a standalone research line.
Back to Research