Charting the Parrot’s Song: A Maximum Mean Discrepancy Approach to Measuring AI Novelty, Originality, and Distinctiveness

2025

Read Charting the Parrot’s Song: A Maximum Mean Discrepancy Approach to Measuring AI Novelty, Originality, and Distinctiveness (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-08446.txt.

Description

Anirban Mukherjee and Hannah Hanwen Chang propose a math test for whether AI-made content is truly new, in a 2025 paper. Courts judge originality by feel, which breaks down when AI systems can produce endless outputs. The pair instead compare whole sets of outputs at once, using a distance measure between the patterns behind two sets.

Their test needs only a handful of samples to tell sets apart. It separated handwritten digit classes with high confidence from five or six samples each. It also split AI art from human art using seven to ten samples per group. Human judges, by contrast, scored barely above chance on the same task. The authors argue that AI outputs form their own distinct pattern rather than simple copies of training data. They offer the method as a tool for courts and policy makers.

Catalog record

YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE174 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-08446

Browse the full library · About the Tower of Babel