Towards Uncertainty-Aware Language Agent
This work addresses uncertainty in language agents for AI and NLP applications, representing an incremental improvement over existing methods like ReAct.
The authors tackled the problem of language agents neglecting uncertainty during interactions with the external world, and introduced the Uncertainty-Aware Language Agent (UALA) framework, which improved performance across tasks like HotpotQA, StrategyQA, and MMLU while reducing reliance on external tools and tokens.
While Language Agents have achieved promising success by placing Large Language Models at the core of a more versatile design that dynamically interacts with the external world, the existing approaches neglect the notion of uncertainty during these interactions. We present the Uncertainty-Aware Language Agent (UALA), a framework that orchestrates the interaction between the agent and the external world using uncertainty quantification. Compared with other well-known counterparts like ReAct, our extensive experiments across 3 representative tasks (HotpotQA, StrategyQA, MMLU) and various LLM sizes demonstrate that UALA brings a significant improvement of performance, while having a substantially lower reliance on the external world (i.e., reduced number of tool calls and tokens). Our analyses provide various insights including the great potential of UALA compared with agent fine-tuning, and underscore the unreliability of verbalised confidence of LLMs as a proxy for uncertainty.