AGITB: A Signal-Level Benchmark for Evaluating Artificial General Intelligence

Matej Šprogar [email protected] · 2025

Read AGITB: A Signal-Level Benchmark for Evaluating Artificial General Intelligence by Matej Šprogar [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-2504-04430.txt.

Description

Matej Šprogar sets out base requirements for artificial general intelligence in this 2025 paper. His premise is that high level intelligence grows from low level signal work: learning by predicting patterns in signals spread across space and time, starting with no grasp of what they mean.

The requirements gather basic principles seen in brain science, from learning in an uninformed state to keeping up in real time. To make the claims testable, and open to being proven wrong, Šprogar builds a testbed with one transparent test per requirement. No non-intelligent system has yet been reported to pass the whole set. He offers the testbed, with public reference code, as a candidate milestone on the road to general intelligence.

Catalog record

CREATORMatej Šprogar [email protected]
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE40 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-04430

Browse the full library · About the Tower of Babel