Learning to Ask: When LLMs Meet Unclear Instruction

2024-08-31

arXiv.org

AI Summary

In real-world applications, user instructions are often ambiguous, causing large language models (LLMs) to blindly hallucinate or over-complete tool parameters—undermining reliability in tool-augmented reasoning. To address this, we propose the Ask-when-Needed (AwN) framework, enabling LLMs to proactively solicit clarifications from users when encountering ambiguities, thereby enhancing robustness in tool invocation under noisy instructions. We introduce NoisyToolBench—the first benchmark for evaluating LLMs on noisy, ambiguous tool-use instructions—and design ToolEvaluator, an automated evaluation suite that jointly measures functional accuracy and interaction efficiency. Furthermore, we identify and characterize a fundamental parameter-hallucination mechanism rooted in LLMs’ next-token prediction bias. Extensive experiments show AwN significantly outperforms state-of-the-art baselines on NoisyToolBench. All code and datasets are publicly released to advance research on robust human–AI collaboration.

Technology Category

• Natural Language Processing: Safety and Robustness
• Machine Learning: Large Multimodal Models (LMMs)
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Equipped with the capability to call functions, modern large language models (LLMs) can leverage external tools for addressing a range of tasks unattainable through language skills alone. However, the effective execution of these tools relies heavily not just on the advanced capabilities of LLMs but also on precise user instructions, which often cannot be ensured in the real world. To evaluate the performance of LLMs tool-use under imperfect instructions, we meticulously examine the real-world instructions queried from users, analyze the error patterns, and build a challenging tool-use benchmark called Noisy ToolBench (NoisyToolBench). We find that due to the next-token prediction training objective, LLMs tend to arbitrarily generate the missed argument, which may lead to hallucinations and risks. To address this issue, we propose a novel framework, Ask-when-Needed (AwN), which prompts LLMs to ask questions to users whenever they encounter obstacles due to unclear instructions. Moreover, to reduce the manual labor involved in user-LLM interaction and assess LLMs performance in tool utilization from both accuracy and efficiency perspectives, we design an automated evaluation tool named ToolEvaluator. Our experiments demonstrate that the AwN significantly outperforms existing frameworks for tool learning in the NoisyToolBench. We will release all related code and datasets to support future research.

Problem

• LLMs struggle with unclear user instructions
• Next-token prediction causes hallucination risks
• Proposed framework improves LLMs' tool-use accuracy

Innovation

• LLMs ask questions on unclear instructions
• Automated tool evaluation enhances efficiency
• NoisyToolBench challenges real-world tool-use scenarios