AutoToM: Scaling Model-based Mental Inference via Automated Agent Modeling
Zhining Zhang 1, ∗ *, Chuanyang Jin 2, ∗ * , † \dagger, Mung Yao Jia 2, ∗ *, Shunchi Zhang 2, ∗ *, Tianmin Shu 2 1 Peking University 2 Johns Hopkins University Link: Project Page | Code · 2025
Read AutoToM: Scaling Model-based Mental Inference via Automated Agent Modeling by Zhining Zhang 1, ∗ *, Chuanyang Jin 2, ∗ * , † \dagger, Mung Yao Jia 2, ∗ *, Shunchi Zhang 2, ∗ *, Tianmin Shu 2 1 Peking University 2 Johns Hopkins University Link: Project Page | Code (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-2502-15676.txt.
Description
Zhining Zhang and co-authors present AutoToM in 2025, a method that helps machines guess what people think and want. Reading minds from acts is key for social machines that work with people. Prompting large models can work but brings steady errors, while hand-built models are firm yet narrow.
AutoToM builds its own model of the person in each task, then reasons backward from acts to beliefs. When doubt runs high, it adds more inner states or more steps of context and tries again. Across five test sets it beat older methods and even large reasoning models. It also gave trust scores much like human ones and aided live choices for a robot.
Catalog record
| CREATOR | Zhining Zhang 1, ∗ *, Chuanyang Jin 2, ∗ * , † \dagger, Mung Yao Jia 2, ∗ *, Shunchi Zhang 2, ∗ *, Tianmin Shu 2 1 Peking University 2 Johns Hopkins University Link: Project Page | Code |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 90 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-15676