Efficient Non-Parametric Uncertainty Quantification for Black-Box Large Language Models and Decision Planning

TL;DR

This paper introduces a non-parametric, efficient uncertainty quantification method for black-box large language models, enabling trustworthy decision planning without extensive computational costs.

Contribution

It presents a novel non-parametric uncertainty estimator for black-box LLMs and a systematic decision-making agent design that improves AI trustworthiness and efficiency.

Findings

Efficient uncertainty estimation with a single inference per input

Improved decision trustworthiness through statistical interpretation

Cost-effective AI agent development without access to token logits

Abstract

Step-by-step decision planning with large language models (LLMs) is gaining attention in AI agent development. This paper focuses on decision planning with uncertainty estimation to address the hallucination problem in language models. Existing approaches are either white-box or computationally demanding, limiting use of black-box proprietary LLMs within budgets. The paper's first contribution is a non-parametric uncertainty quantification method for LLMs, efficiently estimating point-wise dependencies between input-decision on the fly with a single inference, without access to token logits. This estimator informs the statistical interpretation of decision trustworthiness. The second contribution outlines a systematic design for a decision-making agent, generating actions like ``turn on the bathroom light'' based on user prompts such as ``take a bath''. Users will be asked to provide preferences when more than one action has high estimated point-wise dependencies. In conclusion, our uncertainty estimation and decision-making agent design offer a cost-efficient approach for AI agent development.