Summary of A Notion Of Complexity For Theory Of Mind Via Discrete World Models, by X. Angelo Huang et al.
A Notion of Complexity for Theory of Mind via Discrete World Models
by X. Angelo Huang, Emanuele La Malfa, Samuele Marro, Andrea Asperti, Anthony Cohn, Michael Wooldridge
First submitted to arxiv on: 16 Jun 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: Computation and Language (cs.CL); 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 This research proposes a new framework for assessing the complexity of Theory of Mind (ToM) tasks, which are used to evaluate the social reasoning capabilities of Large Language Models (LLMs). The proposed method is inspired by cognitive load theory and quantifies a problem’s complexity as the number of states necessary to solve it correctly. This approach also accounts for spurious states designed to make a task appear harder than it actually is. The authors use their framework to evaluate the complexity of five widely used ToM benchmarks, and demonstrate how a prompting technique called Discrete World Models (DWM) can be used to elicit superior performance on these tasks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps computers understand people’s thoughts by making them solve complex puzzles that require social reasoning. The researchers created a new way to measure how hard these puzzles are based on the number of possible solutions. They also developed a special way to give hints to the computer about how the puzzle changes as it’s solved. This helped the computer perform better at solving the puzzles. |
Keywords
» Artificial intelligence » Prompting
