Prompt Risk Control: A Rigorous Framework for Responsible Deployment of Large Language Models
This work addresses the need for responsible deployment of large language models by mitigating risks for users, though it is incremental as it builds on existing statistical bounding techniques.
The authors tackled the problem of selecting prompts for large language models to avoid unexpectedly poor responses, especially for worst-off users, by proposing Prompt Risk Control, a framework that provides rigorous upper bounds on risk measures and reduces the risk of worst outcomes in applications like chat and code generation.
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.