How Did You Do On The AI Art Turing Test?

Results of Scott's AI-art-vs-human-art identification test (11,000 participants, 50 curated images). Findings: median score 60%, barely above chance — most people can't spot AI art on subtle style/quality alone; respondents judged by style (clumping all Impressionism as human); they slightly preferred AI art (the top two favorites were AI); even self-described AI-art-loathers preferred it blind; but genuinely skilled viewers (artists, AI-haters) scored higher, illustrated by his friend 'Ilzo's' detailed dissection of why an AI gateway is 'superficially detailed but incoherent' and the key tell that 'human artists get vague when they don't know the details' while AI is confidently wrong. Concludes that AI passes the art Turing Test, that humans venerate prestige and dismiss the new/low-status without engaging the work, and (Duchamp-urinal callback) that 'Sam Altman is the greatest artist of the 21st century.'

Quality 74 · Strong. Strong band, upper. Genuinely informative empirical findings (people can't tell, prefer AI, judge by style and prestige) plus a real analytical nugget about AI's failure mode (confident incoherence vs human strategic vagueness). Data-plus-commentary rather than a single thesis, so upper-Strong.

Claude’s paradigm shift 45 · Moderate. Moderate. The experiment yields novel-ish results and the prestige-bias and confident-incoherence observations are fresh, but the underlying ideas (Turing test, status-driven aesthetic judgment) pre-exist.

Real-world impact 2 · Minor. Genuinely informative empirical findings from his 11,000-participant AI-art Turing test (people can't tell, slightly prefer AI, judge by style/prestige) plus a real nugget on AI's failure mode (confident incoherence vs human strategic vagueness). Real topical relevance to the AI-art debate, data-plus-commentary, no material change — low RWI.