Adaptive Tool Use in Large Language Models with Meta-Cognition Trigger
This addresses efficiency and reliability issues in LLM tool use for AI applications, but it is incremental as it builds on existing tool-access research.
The paper tackles the problem of indiscriminate tool invocation in large language models (LLMs), which increases latency and errors, by introducing MeCo, an adaptive decision-making strategy based on meta-cognition that quantifies cognitive signals to guide tool use; experiments show it significantly improves tool-use decision-making across multiple models and benchmarks.
Large language models (LLMs) have shown remarkable emergent capabilities, transforming the execution of functional tasks by leveraging external tools for complex problems that require specialized processing or up-to-date data. While existing research expands LLMs access to diverse tools (e.g., program interpreters, search engines, calculators), the necessity of using these tools is often overlooked, leading to indiscriminate tool invocation. This naive approach raises two key issues: increased latency due to unnecessary tool calls, and potential errors resulting from faulty interactions with external tools. In this paper, we introduce meta-cognition as a proxy for LLMs self-assessment of their capabilities, reflecting the model's awareness of its own limitations. Based on this, we propose MeCo, an adaptive decision-making strategy for external tool use. MeCo quantifies metacognitive scores by capturing high-level cognitive signals in the representation space, guiding when to invoke tools. Notably, MeCo is fine-tuning-free and incurs minimal cost. Experiments across multiple backbone models and benchmarks show that MeCo reliably detects LLMs' internal cognitive signals and significantly improves tool-use decision-making.