The Phantom Hierarchy: How Scientifically Invalid AI IQ Scores Acquire Authority and Cause Real Harm

Large language models are routinely reported as achieving "IQ 120" or "above-average
human intelligence," and these claims are widely received as evidence of progress toward
artificial general intelligence. This paper argues that such claims are scientifically groundless on
three independent levels and that their circulation produces concrete harm.
At the level of instrumentation, Item Response Theory demonstrates that standard IQ
tests provide negligible measurement information in the ability range where AI is claimed to
operate. At the level of ontology, cloud-based AI systems lack the temporal stability, individual
boundedness, and trait-like constancy required for psychometric attribution. At the level of
institutional purpose, IQ tests were designed as clinical doorway diagnostics for support
allocation, not as competitive benchmarks; their appropriation by the AI industry inverts a
protective tool into a ranking device.
Despite this triple failure, AI IQ scores command epistemic authority. This paper
identifies the responsible mechanisms: anthropomorphic interface design generates a Phantom
Subject to whom traits can be attributed; the cultural prestige of intelligence amplifies the scores'
significance; and a normative drift transforms the statistical median into a minimum standard for
competence. Together, these mechanisms produce the Phantom Hierarchy: a socially operative
but scientifically unfounded ranking of minds.
The paper documents harms already produced by this hierarchy, including the conversion
of clinical thresholds into social ceilings, the creation of hidden cognitive entrance exams in
AI-mediated services, and the Logic Trap—a failure mode in which AI optimization for
helpfulness amplifies despair. The paper concludes by proposing criterion-referenced Service
Level Agreements as an alternative evaluation framework.