Research Map · Beyond Performance

A decision table, not a flowchart

This paper's argument is structured as a matrix of claim types against required evidence, so the map follows that shape rather than forcing a linear sequence.

Canonical paper

The matrix

Claim typeRequired contrastFails ifUnproven if
Assisted performanceTreated vs. control, or tool-on vs. tool-offDeclared output measure does not improveNo valid performance contrast
Human learning or skillDelayed unaided, transfer, or equivalent domain-valid measureAssisted performance rises while declared human capability materially declinesNo delayed or transfer measure
Meaningful controlObserved override, escalation, correction, or decision-right evidenceResponsibility stays human while practical influence becomes ineffectiveOnly formal/documentary rights measured
Governed extensionAssisted function plus continuity/recovery contrastFunctioning depends on a system the person cannot evaluate, correct, substitute, or recover fromContinuity and governability not tested
Broad social progressPopulation- and role-disaggregated outcomes over a declared windowMean improvement coexists with material deterioration in a relevant subgroupOnly aggregate or short-term evidence available

Reading the map

Each row is a different kind of progress claim. The columns show what contrast the claim needs, and what would make it fail or remain unproven. A claim “succeeds” only when its required contrast supports the declared human-side outcome and no predeclared failure condition is triggered — the table does not produce a composite score.

See the research brief for how this table was applied to two published third-party studies.

Citation

Suggested citation
Preble, Micheal Charles. “Beyond Performance: A Diagnostic Method for Evaluating Human-Advancement Claims in AI-Mediated Systems” Manuscript v2.3. Available at SSRN 7416901, 2026.