Prompt Risk Control: A Rigorous Framework for Responsible Deployment of Large Language Models

TL;DR

This paper introduces Prompt Risk Control, a framework that uses rigorous statistical bounds to select prompts for large language models, aiming to minimize worst-case responses and disparities, thereby promoting responsible deployment.

Contribution

It presents a novel framework for prompt selection based on upper bounds of risk measures, including worst-case and distribution-shifted scenarios, enhancing safety in LLM deployment.

Findings

Reduces worst-case response risks in LLM applications

Balances generation quality across diverse user groups

Effective under distribution shifts in deployment environments

Abstract

The recent explosion in the capabilities of large language models has led to a wave of interest in how best to prompt a model to perform a given task. While it may be tempting to simply choose a prompt based on average performance on a validation set, this can lead to a deployment where unexpectedly poor responses are generated, especially for the worst-off users. To mitigate this prospect, we propose Prompt Risk Control, a lightweight framework for selecting a prompt based on rigorous upper bounds on families of informative risk measures. We offer methods for producing bounds on a diverse set of metrics, including quantities that measure worst-case responses and disparities in generation quality across the population of users. In addition, we extend the underlying statistical bounding techniques to accommodate the possibility of distribution shifts in deployment. Experiments on applications such as open-ended chat, medical question summarization, and code generation highlight how such a framework can foster responsible deployment by reducing the risk of the worst outcomes.