Summary of Llms Can’t Plan, but Can Help Planning in Llm-modulo Frameworks, by Subbarao Kambhampati et al.
LLMs Can’t Plan, But Can Help Planning in LLM-Modulo Frameworks
by Subbarao Kambhampati, Karthik Valmeekam, Lin Guan, Mudit Verma, Kaya Stechly, Siddhant Bhambri, Lucas Saldyt, Anil Murthy
First submitted to arxiv on: 2 Feb 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: 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 paper aims to clarify the role of Large Language Models (LLMs) in planning and reasoning tasks. While some claim that LLMs can perform these tasks with suitable prompting or self-verification strategies, others argue that they are merely translators, passing problems to external symbolic solvers. The authors propose a more nuanced view, arguing that auto-regressive LLMs cannot handle planning or self-verification alone, but can be used as universal approximate knowledge sources. They present the concept of LLM-Modulo Frameworks, which combines the strengths of LLMs with external model-based verifiers in a bi-directional interaction regime. This framework enables the acquisition of models driving external verifiers with the help of LLMs. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Large Language Models are very smart computers that can understand and generate human-like text. Some people think they’re great at planning and making decisions, but others say they’re not so good at this. The authors of a new paper want to clear up some confusion about how these models work. They think that the models can’t do all the planning and decision-making on their own, but they can help us get better ideas by giving us information we need. The authors propose a way to use these models together with other computer programs to make even better decisions. |
Keywords
* Artificial intelligence * Prompting
