Time-Scaling Is What Agents Need Now

Zhi Liu, Guangzhi Wang · 2026

Read Time-Scaling Is What Agents Need Now by Zhi Liu, Guangzhi Wang (2026) 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-2601-02714.txt.

Description

Zhi Liu and Guangzhi Wang argue that what modern AI agents most lack is time scaling, the ability to spend sequential reasoning time well. Early AI split cognition apart, neural networks for perception, reinforcement learning for decisions, and symbolic systems for knowledge. But transformer based large models and world models are merging those strands into agents that perceive, decide, and act in a loop. Humans, with limited resources, solve hard problems by reasoning step by step in time.

Prompting methods that externalize intermediate steps, chain of thought and tree of thought, lengthened the reasoning path and helped, and recent models trained with explicit reasoning traces went further. The authors find those methods still limited in how completely and efficiently they search. Their position is that scaling the time dimension of reasoning, deliberately and structurally, is the next lever for agent capability.

Catalog record

CREATORZhi Liu, Guangzhi Wang
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE46 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2601-02714

Browse the full library · About the Tower of Babel