The program's foundational assumption, published as a falsifiable object rather than embedded in a purpose statement. Of the three possible bridge claims — resemblance, influence, replacement — this hypothesis commits to influence as the primary channel: AI evaluators shape what humans see, shortlist, and trust, regardless of whether their internal evaluation resembles a committee's.
We deliberately do not claim resemblance. Evaluator-stated rationales are observations about the evaluator's narration, not evidence of mechanism (see C-0002), and nothing yet licenses the stronger claim.
Predictions — registered before results
- Buying decisions in AI-screened categories will show measurable dependence on evaluator recommendation behavior
- Changes to a vendor's machine-legible representation will propagate to human shortlists via AI-assisted screening
What would refute this
- Longitudinal evidence that AI evaluator outputs have no measurable effect on human shortlist composition or selection
- Evidence that human committee evaluation and AI evaluation diverge so completely that findings about one predict nothing about the other