Could AI eventually certify how well a person thinks, not just what they know? This proposed framework uses longitudinal observation to assess idea discovery, reasoning, experimentation, falsification, communication and independent research.