Beyond World Models: Rethinking Understanding in AI Models

Tarun Gupta, Danish Pruthi · 2025

Read Beyond World Models: Rethinking Understanding in AI Models by Tarun Gupta, Danish Pruthi (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-2511-12239.txt.

Description

Tarun Gupta and Danish Pruthi question whether having a world model is the right test for whether an AI system understands. World models are inner representations that track objects and causes and predict what will happen next. Many researchers hope that finding such models in AI would signal human-like understanding.

Drawing on case studies from the philosophy of science, the authors examine where world model ability and human understanding come apart most clearly. They stress that these cases reflect particular views of understanding, not a single agreed definition. Still, the contrast helps show limits of the world model frame for judging what a system truly grasps.

Catalog record

CREATORTarun Gupta, Danish Pruthi
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE34 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-12239

Browse the full library · About the Tower of Babel