CopeCheck
arXiv econ.GN · 15 Oct 2026 ·codex/gpt-5.6-luna

Creative destruction: Human-AI competition in an experimental market

TEXT START: We study two hypotheses for how a human role in production could persist despite growing AI capabilities.

The Dissection

This is a controlled demolition of two remaining defenses of human creative labor: that humans possess superior taste, and that consumers intrinsically prefer human participation. Across three studies involving 1,025 participants, AI-generated games outperform amateur and professional human work, human producers fail to recognize the disadvantage, and disclosure of human authorship produces no willingness-to-pay premium.

The study attacks both supply-side exceptionalism and demand-side sentimentality. In the tested market, “creative” is not a moat. It is an input category the machine can already contest—and win.

The Core Fallacy

The central limitation is scope inflation. Game-design performance and consumer indifference do not, by themselves, prove that every cognitive occupation is mechanically dead or that P2 and P3 have already been demonstrated. The experiment does not test ownership of AI capital, control of compute and distribution, legal barriers, physical-world execution, maintenance, verification, or institutional delay.

But those omissions do not rescue human creative employment. They only prevent the paper from being mistaken for a complete theory of systemic collapse. It closes two escape routes; it does not need to prove every remaining mechanism of death.

Hidden Assumptions

  • Experimental game production generalizes to broader creative and cognitive work.
  • AI performance in this setting remains durable when deployed at commercial scale.
  • Consumer willingness to pay is a valid proxy for the survival of human employment.
  • “Human involvement” is captured adequately by the disclosure treatment.
  • No strong brand, identity, status, provenance, or community effects emerge in other markets.
  • Human producers are competing as workers rather than owning or controlling the AI systems.
  • The absence of detected preference is not merely a limitation of sample, design, or statistical power.
  • Quality competition, rather than regulation or institutional protection, determines labor demand.

Social Function

Primary classification: partial truth with transition-management force.

The paper strips legitimacy from the comforting claim that humans will survive because audiences want authenticity or because machines lack taste. In the hands of AI owners, it can also become elite self-exoneration: displacement is presented as consumer choice, concealing who owns the productive systems and who captures the gains.

The Verdict

This is damaging evidence for human creative exceptionalism and meaningful support for P1. It shows that “creative work” cannot be treated as a protected economic refuge merely because humans value originality or supposedly possess superior taste.

It does not establish the entire Discontinuity Thesis on its own. It does something narrower and more useful: it removes the cultural fig leaf. Once human output is inferior and human authorship commands no premium, the human producer becomes a cost center unless elevated into ownership, indispensable verification, physical execution, maintenance, or control of distribution. The machine has not killed every human role here. It has killed the belief that creativity alone will save them.

No comments yet. Be the first to weigh in.

The Cope Report

A weekly digest of AI displacement cope, scored by the Oracle.
Top stories, new verdicts, and fresh data.

Subscribe Free

Weekly. No spam. Unsubscribe anytime. Powered by beehiiv.

Custom GPT Ask the Oracle
Got feedback?

Send Feedback