Summary of Muma-tom: Multi-modal Multi-agent Theory Of Mind, by Haojun Shi et al.
MuMA-ToM: Multi-modal Multi-Agent Theory of Mind
by Haojun Shi, Suyu Ye, Xinyu Fang, Chuanyang Jin, Leyla Isik, Yen-Ling Kuo, Tianmin Shu
First submitted to arxiv on: 22 Aug 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper introduces MuMA-ToM, a novel benchmark for understanding people’s mental states in complex real-world scenarios, specifically focusing on Theory of Mind reasoning in multi-agent interactions. This framework evaluates AI systems’ ability to infer mental states from multi-modal information, such as video and text descriptions of social interactions. The authors provide a human baseline through experiments and propose LIMP, a language model-based inverse multi-agent planning algorithm that outperforms state-of-the-art methods, including large multi-modal models like GPT-4o and Gemini-1.5 Pro. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine trying to understand what people are thinking when they interact with each other. This paper develops a new way to test artificial intelligence (AI) systems’ ability to do just that. The approach uses videos and text descriptions of people’s behaviors in everyday situations, like a household, to help AI systems figure out what people want and believe about others. The researchers also created a special algorithm, LIMP, which performs better than current AI models at understanding these mental states. |
Keywords
» Artificial intelligence » Gemini » Gpt » Language model » Multi modal
