A research problem
Can AI improve your performance without improving your capability?
A model can become more capable while a person's assisted output rises, retained capacity falls, relationship state becomes more consequential, or practical authority shifts away from the person. A gain in one layer does not establish a gain in another.
Most performance measures record the gain and stop there. This research asks whether an AI-assisted gain became durable human capacity, or whether the improvement disappears once the assistance is removed, delayed, or changed.
Relevant research
- Measurement — the diagnostic framework separating model capability, assisted performance, human capacity, relationship state, and practical authority.
- Beyond Performance — a pre-validation method for deciding whether improved AI-mediated performance supports a claim about human advancement, remains unproven, or fails.
- The Human-Development Conversion Constraint — what evidence is required before increasing artificial capability can support a broader claim about human progress.
- Human Capacity Protection (HCP) — asks whether apparent gains during AI-supported activity survive transfer, delay, removal, or changed conditions.