A Use-Case Specific Dataset for Measuring Dimensions of Responsible Performance in LLM-generated Text
This provides a concrete resource for the research community to evaluate LLMs on Responsible AI dimensions in a targeted application, addressing a gap in current high-level evaluation methods.
The authors tackled the problem of evaluating large language models (LLMs) for Responsible AI dimensions like fairness by constructing a use-case specific dataset for generating product descriptions, parameterized by fairness attributes, gendered adjectives, and product categories, and showed how to use it to identify quality, veracity, safety, and fairness gaps in LLMs.
Current methods for evaluating large language models (LLMs) typically focus on high-level tasks such as text generation, without targeting a particular AI application. This approach is not sufficient for evaluating LLMs for Responsible AI dimensions like fairness, since protected attributes that are highly relevant in one application may be less relevant in another. In this work, we construct a dataset that is driven by a real-world application (generate a plain-text product description, given a list of product features), parameterized by fairness attributes intersected with gendered adjectives and product categories, yielding a rich set of labeled prompts. We show how to use the data to identify quality, veracity, safety, and fairness gaps in LLMs, contributing a proposal for LLM evaluation paired with a concrete resource for the research community.