Summary of Comma: a Communicative Multimodal Multi-agent Benchmark, by Timothy Ossowski et al.
COMMA: A Communicative Multimodal Multi-Agent Benchmark
by Timothy Ossowski, Jixuan Chen, Danyal Maqbool, Zefan Cai, Tyler Bradshaw, Junjie Hu
First submitted to arxiv on: 10 Oct 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: None
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 Multimodal agents built on large foundation models have made significant strides in various tasks, but there’s a critical gap in understanding their language-based communication abilities between agents in collaborative settings. Existing benchmarks fall short of evaluating inter-agent communication and collaboration effectively, particularly when agents have unequal access to information and must work together to achieve complex tasks. To address this gap, we introduce a novel benchmark designed to evaluate the collaborative performance of multimodal multi-agent systems through language communication. Our benchmark features various scenarios, providing comprehensive evaluation across four key categories of agentic capability in a communicative collaboration setting. We test both agent-agent and agent-human collaborations using open-source and closed-source models, revealing surprising weaknesses in state-of-the-art models, including proprietary models like GPT-4o. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Language-based communication between agents is crucial for collaborative tasks, but existing benchmarks don’t evaluate it effectively. A new benchmark fills this gap by testing how well multimodal multi-agent systems work together through language. Our test features different scenarios and shows how well agents can collaborate when some have more information than others. We tested popular models like GPT-4o and found that even the best ones struggle to work together unless a human is involved. |
Keywords
» Artificial intelligence » Gpt
