Summary of Evaluating Language Model Agency Through Negotiations, by Tim R. Davidson et al.
Evaluating Language Model Agency through Negotiations
by Tim R. Davidson, Veniamin Veselovsky, Martin Josifoski, Maxime Peyrard, Antoine Bosselut, Michal Kosinski, Robert West
First submitted to arxiv on: 9 Jan 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); 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 proposed approach evaluates language model agency using negotiation games, addressing shortcomings of existing benchmarks. This method enables studying multi-turn interactions, modulating complexity, and preventing accidental evaluation data leakage. Six widely used LMs are tested in self-play and cross-play settings, revealing insights such as only closed-source models completing tasks, cooperative bargaining games being challenging, and even powerful models sometimes “losing” to weaker opponents. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A team of researchers developed a new way to test how well language models work together. They used special games that involve negotiation and decision-making. This helps them see how the models perform in real-life situations. The team tested six popular language models and found some interesting things. For example, only certain models that aren’t publicly available were able to complete these tasks on their own. They also discovered that when models had to work together, it was really hard for them to make good decisions. Sometimes, the stronger models even lost to weaker ones. |
Keywords
* Artificial intelligence * Language model
