Towards Measurement Theory for Artificial Intelligence (Extended Abstract)

Elija Perrier [email protected] · 2025

Read Towards Measurement Theory for Artificial Intelligence (Extended Abstract) by Elija Perrier [email protected] (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-2507-05587.txt.

Description

Elija Perrier argues that artificial intelligence needs a proper theory of measurement, and sketches one. The extended abstract appeared in twenty twenty-five. Fields from engineering learned long ago that a measure is itself a choice, shaping what a capability claim even means.

Formalizing measurement would let researchers compare systems and the tests applied to them on common ground. It would also connect frontier AI evaluation to the quantitative risk tools of safety science. Perrier outlines a layered measurement stack and separates direct observables from indirect ones. The sketch points toward a shared, calibratable taxonomy of AI phenomena. Regulators, builders, and evaluators are the intended audience for the programme.

Catalog record

CREATORElija Perrier [email protected]
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE51 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-05587

Browse the full library · About the Tower of Babel