What problem does it solve? Texts about AI often blur the line between describing what a system does (causal, functional, or computational activity) and claiming it has intentions, consciousness, understanding, desires, or autonomous will. This Skill audits writing to detect where human-like mental properties are attributed to AI or technological systems without sufficient conceptual or textual justification, while avoiding false alarms on legitimate functional language. ## Core Features & Use Cases - Mental Attribution Detection: Identifies explicit or implicit claims of consciousness, intention, desire, belief, understanding, will, or autonomous agency in AI descriptions. - Functional vs. Mental Distinction: Classifies each attribution as causal, functional, relational, computational, intentional, epistemic, or ambiguous, so ordinary active verbs like "generates" or "transforms" are not wrongly flagged. - Structured Findings Report: Outputs findings with evidence quotes, severity classification (A–E), confidence level, implications, and revision recommendations. - Use Case: A philosophy researcher reviewing a manuscript on AI agency can run this audit to catch passages where "the system operates autonomously" silently drifts into "the system pursues its own goals" without an intervening argument. ## Quick Start Audit the attached essay on AI decision-making for unjustified anthropomorphic attributions and report each finding with its classification and recommendation.
