Executive Brief · Human-Development Conversion Constraint
What institutions should measure before calling AI adoption progress
Micheal Charles Preble · Independent Researcher · For policy and organizational readers
Why this reaches decision-makers
Institutions routinely cite AI adoption, productivity gains, or system capability as evidence of progress for the people they serve. This paper's constraint bears directly on that inference, and several of its objections-and-responses (Sections 7, 10.2, 10.6, 10.7) speak specifically to institutional and market contexts rather than to the philosophical claim alone.
The decision-relevant distinction
Adoption, productivity, and market success are evidence about technology and organizations — not, by themselves, evidence about the people an institution serves. Two institutions can report identical AI adoption rates while producing very different human consequences, depending on distribution and durability.
What the paper says about markets
Market adoption reveals willingness to use a technology under existing constraints; it does not establish that use produces durable capability, that benefits are broadly distributed, or that people become dependent on systems that later affect their practical freedom. Market success and human-development success are related but not interchangeable measures.
What the paper says about metrics and gaming
If institutions begin evaluating AI adoption through a “human-side outcome” metric, that metric can itself be optimized without improving the underlying condition it was meant to track (a Goodhart effect). The constraint is intentionally decomposed into multiple distinct questions rather than compressed into one composite score, specifically to resist this failure mode.
What the paper says about distribution and positional goods
A broadly distributed capability gain does not necessarily translate into better access to scarce, positional goods — selective education, scarce employment, institutional power. An institution may report rising average capability while institutional power becomes more concentrated. The constraint requires that this divergence be exposed, not that it be resolved by a single policy.
Evidence Status
Evidence status: Conceptual, philosophical constraint. This brief translates its institutional implications; it does not add new empirical claims.
What This Does Not Claim
- Does not prescribe a specific tax, procurement rule, or institutional mechanism.
- Does not claim distribution has been overlooked in AI scholarship generally — its narrower claim is that distribution should condition the inference from technological benefit to broad human progress.
- Does not resolve the objection that causal attribution may be difficult; it constrains methodology rather than demanding certainty research cannot supply.
Read / Cite the Research
Read the canonical paper record → · Full paper PDF
Preble, Micheal Charles. “The Human-Development Conversion Constraint: What Should Count as Progress in AI-Rich Societies?” Available at SSRN 7416900, 2026.