LLM as a Mastermind: A Survey of Strategic Reasoning with Large Language Models

This paper presents a comprehensive survey of the current status and opportunities for Large Language Models (LLMs) in strategic reasoning, a sophisticated form of reasoning that necessitates understanding and predicting adversary actions in multi-agent settings while adjusting strategies accordingly. Strategic reasoning is distinguished by its focus on the dynamic and uncertain nature of interactions among multi-agents, where comprehending the environment and anticipating the behavior of others is crucial. We explore the scopes, applications, methodologies, and evaluation metrics related to strategic reasoning with LLMs, highlighting the burgeoning development in this area and the interdisciplinary approaches enhancing their decision-making performance. It aims to systematize and clarify the scattered literature on this subject, providing a systematic review that underscores the importance of strategic reasoning as a critical cognitive capability and offers insights into future research directions and potential improvements.

Paper

References (92)

06Human vs. Machine: Language Models and Wargames2024

Scroll for more · 38 remaining

Similar papers

Reviewer dKAH6/10 · confidence 4/52024-05-08

Summary

This paper provides a comprehensive survey about the field of strategic reasoning with LLMs. The authors give a clear definition of strategic reasoning, and core characteristics compared to other reasoning types. They categorize and systematically review the scenarios applying strategic reasoning with LLMs, existing methods enhancing the LLMs' capability for strategic reasoning, and the evaluation in strategic reasoning with LLMs. The challenges of strategic reasoning with LLMs are discussed in the end.

Rating

6

Confidence

4

Ethics flag

1

Reasons to accept

1. This is a timely survey for strategic reasoning with LLMs, which is clearly an important research topic. 2. The survey is well-organized and clearly written. I especially like the formalism of the definition of strategic reasoning. 3. A broad range of works related to strategic reasoning with LLMs are well placed.

Reasons to reject

1. The categorization of scenarios only distinguishes topic differences and does not discuss how they uniquely test the reasoning capabilities of LLMs or how they present distinct features and challenges compared to other scenarios. 2. In the first few sections, the survey could benefit from a deeper analysis of the related works rather than just listing them. It would be valuable to see a more detailed examination that connects different studies, evaluates their contributions critically, and outlines a developmental roadmap of the research area. This is essential for readers to grasp the progression and interplay of the works within the field. 3. While the discussion and invitation for future contributions from the community is appreciated and well-timed, it would be beneficial to include the authors' perspectives on practical approaches to realizing them.

Questions to authors

Typos: - Mind the space before citations. - public debatesCollectively

Reviewer 3S3R7/10 · confidence 4/52024-05-09

Summary

This survey analyzes the problem definition, solutions, and evaluation of LLM strategic reasoning from various perspectives. It also provides future directions for research in the field of LLM strategic reasoning. I really like this paper and recommend it to be accepted.

Rating

7

Confidence

4

Ethics flag

1

Reasons to accept

1. This paper is clearly written and easy to follow. 2. The perspective proposed in this paper is excellent. Strategic reasoning is a crucial issue within LLMs, serving as an important dimension for evaluating LLM capabilities and a significant indicator for achieving AGI 3. The classification basis for the method of strategic reasoning proposed in this article is very reasonable, and the literature review is conducted with great thoroughness. 4. In the discussion of future work, the authors note that we still lack studies on the boundaries of strategic reasoning capabilities in large models, which is indeed a crucial point.

Reasons to reject

1. A minor critique is the lack of discussion on the conditions under which quantitative and qualitative evaluation metrics are applicable when analyzing the evaluation, specifically why these particular metrics were chosen.

Questions to authors

I do not have questions but I recommend adding more related work: [1] TimeArena: Shaping Efficient Multitasking Language Agents in a Time-Aware Simulation [2] AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation [3] SCIENCEWORLD: Is your Agent Smarter than a 5th Grader?

Reviewer 5o6Z6/10 · confidence 3/52024-05-10

Summary

In this paper, the authors give a comprehensive review of recent work on using LLMs in strategic reasoning environments (e.g., gaming environments, economic simulation). This area of research has gained much traction in the last year (e.g., all of the papers they cite in Figures 2 and 4 involves work done exclusively in 2023 and 2024) and is integral to new applications of LLMs in neighboring disciplines such as economics, business and many others. I therefore think that a survey paper of this kind is very timely and important and I generally found their paper to be easy to read and accessible. As with any broad survey article, certain liberties were taken with how to classify different work and concepts. For example, they exclude generative agents [Park et al] and work like this as being out of scope given that the lack of focus on goal-oriented task modeling (I’d like to see more discussion of this). They also attempted to quantify the different types of reasoning skills involved with strategic reasoning tasks versus other tasks and I found this part to be overly subjective without further discussion. Given that virtually all of this work involves prompt-based LLMs, distinguishing between “prompt-engineering” approaches versus other approaches, virtually all of which involve prompting, to be a bit confusing (in fairness, the authors point this out when they write that “it is important to note that the boundaries between the above methodological categories are not entirely orthogonal”.) Given that this article is only a survey article and offers no new empirical or technical results, I don’t have much to criticize outside of some presentation-specific issues that I note below. Baring these concerns, my main concern is about whether a survey paper fits within the COLM technical track (I don’t think it does, I will say more below. I am willing to be convinced otherwise if other reviewers and the area chair(s) are not concerned about this).

Rating

6

Confidence

3

Ethics flag

1

Reasons to accept

- A comprehensive and very easy to read review of LLMs and strategic reasoning, which is motivated by the considerable interest in the topic in the last year. I think it provides a useful overview of this emerging area and that researchers will cite it for this reason.

Reasons to reject

- No new technical or empirical results; not original research but a survey article. Given the call for papers (https://colmweb.org/cfp.html) and the reviewer guidelines (https://colmweb.org/ReviewGuide.html, in particular, the stated goal of having a “technical deep, exciting, forward-looking, insightful and impact program”), I don’t the paper meets the criterion for inclusion in COLM. - (a complicated criticism) While their review is comprehensive, it fails, in my mind, to communicate what the big problems are in the different application areas where strategic reasoning has been investigated. For example, in the scenarios involving economics or game theory, what are the big or open problems in these fields that motivate using language models in place of traditional tools? What are the prospects of success in using LLMs for these problems? Of these problems, are there particular sets of problems that the LLM field has tended to focused on, or ones that people are not paying attention to? A lot of the discussion along these lines is left vague (e.g., they end the “scenarios” section by saying: “each category [or application area of LLMs] offer[s]… unique insights and challenges”. I was really keen to get deeper into these challenges). I think it would benefit the LLM community to know a bit more about what the core problems are in these different to better motivate interest in strategic reasoning. It might also make the article easier for others outside of LLMs to engage with this work.

Reviewer ppn36/10 · confidence 3/52024-05-14

Summary

This paper provides a survey regarding strategic reasoning with LLMs. The paper well motivated why LLMs are game changers for strategic reasoning and frame the existing works according to application, methods, and evaluation. Interesting challenges and insights are discussed at the end.

Rating

6

Confidence

3

Ethics flag

1

Reasons to accept

- Strategic reasoning is an important topic. Though many existing works (especially relying on LLMs) are published recently, their effort indeed widely spread in terms of topics and evaluation criteria. Thus, it becomes hard to follow and compare across papers. The survey paper calls out a unified framework and benchmark, which is crucial and necessary for the related research community. Researcher should align on the same standard. - The survey provides good angles to well unify most of the related works.

Reasons to reject

- As a survey paper, the paper has limited novelty. - Some points on outlook section can go deeper with more interesting insights and designs among existing work.

Questions to authors

I agree with the point that we should have a unified framework and benchmark. I am curious how could we define any proposed benchmark as unified and standard enough so that every related research should evaluate with?

Program Chairsdecision2024-07-10

Decision

Accept

© 2026 NYSGPT2525 LLC