MACI: Multi-Agent Collaborative Intelligence for Adaptive Reasoning and Temporal Planning

Edward Y. Chang, Stanford University · 2025

Read MACI: Multi-Agent Collaborative Intelligence for Adaptive Reasoning and Temporal Planning by Edward Y. Chang, Stanford University (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-2501-16689.txt.

Description

Edward Chang builds MACI, a framework that lets groups of AI agents plan tasks with care for time and limits. His paper from two thousand twenty-five says large language models often plan by pattern alone. They check their own work poorly and handle limits in uneven ways.

MACI splits the job three ways. A meta-planner lists roles and limits, adds common-sense fixes, and draws a graph of task order. A set of agents then does the task work, while a run-time monitor adjusts the plan as things change. By keeping planning apart from checking, MACI did well on two scheduling tasks, such as wedding planning.

Catalog record

CREATOREdward Y. Chang, Stanford University
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE83 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-16689

Browse the full library · About the Tower of Babel