September 20, 2025

Empowering LLMs with Logical Reasoning: A Comprehensive Survey

View Full Paper

Ask AI

Bookmark

Authors

Discussion

Survey reveals challenges in logical reasoning for large language models, indicating needs for improved consistency.

Logical reasoning capabilities are essential for large language models to correctly answer complex problems, particularly in logical question answering.
Instances of logical inconsistency in responses highlight significant shortcomings in current large language models, raising concerns for practical applications.
A comprehensive taxonomy categorizes methods aimed at enhancing logical reasoning by focusing on external solvers, prompts, and fine-tuning.
Extending research to modal logic and developing efficient algorithms may help address challenges related to logical consistencies and uncertainties.

Cheng et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark