Validating LLM-as-a-Judge Systems in the Absence of Gold Labels

The LLM-as-a-judge paradigm, in which a judge LLM system replaces human raters in rating the outputs of other generative AI (GenAI) systems, has come to play a critical role in scaling and standardizing GenAI evaluations. To validate judge systems, evaluators collect multiple human ratings for each item in a validation corpus, and then aggregate the ratings into a single, per-item gold label rating. High agreement rates between these gold labels and judge system ratings are then taken as a sign of good judge system performance. In many cases, however, items or rating criteria may be ambiguous, or there may be principled disagreement among human raters. In such settings, gold labels may not exist for many of the items. In this paper, we introduce a framework for LLM-as-a-judge validation in the absence of gold labels. We present a theoretical analysis drawing connections between different measures of judge system performance under different rating elicitation and aggregation schemes. We also demonstrate empirically that existing validation approaches can select judge systems that are highly suboptimal, performing as much as 34% worse than the systems selected by alternative approaches that we describe. Based on our findings, we provide concrete recommendations for developing more reliable approaches to LLM-as-a-judge validation.

Toward an Evaluation Science for Generative AI Systems

There is an increasing imperative to anticipate and understand the performance and safety of generative AI systems in real-world deployment contexts. However, the current evaluation ecosystem is insufficient: Commonly used static benchmarks face validity challenges, and ad hoc case-by-case audits rarely scale. In this piece, we advocate for maturing an evaluation science for generative AI systems. While generative AI creates unique challenges for system safety engineering and measurement science, the field can draw valuable insights from the development of safety evaluation practices in other fields, including transportation, aerospace, and pharmaceutical engineering. In particular, we present three key lessons: Evaluation metrics must be applicable to real-world performance, metrics must be iteratively refined, and evaluation institutions and norms must be established. Applying these insights, we outline a concrete path toward a more rigorous approach for evaluating generative AI systems.

Machine Unlearning Doesn't Do What You Think: Lessons for Generative AI Policy, Research, and Practice

We articulate fundamental mismatches between technical methods for machine unlearning in Generative AI, and documented aspirations for broader impact that these methods could have for law and policy. These aspirations are both numerous and varied, motivated by issues that pertain to privacy, copyright, safety, and more. For example, unlearning is often invoked as a solution for removing the effects of targeted information from a generative-AI model's parameters, e.g., a particular individual's personal data or in-copyright expression of Spiderman that was included in the model's training data. Unlearning is also proposed as a way to prevent a model from generating targeted types of information in its outputs, e.g., generations that closely resemble a particular individual's data or reflect the concept of "Spiderman." Both of these goals--the targeted removal of information from a model and the targeted suppression of information from a model's outputs--present various technical and substantive challenges. We provide a framework for thinking rigorously about these challenges, which enables us to be clear about why unlearning is not a general-purpose solution for circumscribing generative-AI model behavior in service of broader positive impact. We aim for conceptual clarity and to encourage more thoughtful communication among machine learning (ML), law, and policy experts who seek to develop and apply technical methods for compliance with policy objectives.

A Framework for Evaluating LLMs Under Task Indeterminacy

Large language model (LLM) evaluations often assume there is a single correct response -- a gold label -- for each item in the evaluation corpus. However, some tasks can be ambiguous -- i.e., they provide insufficient information to identify a unique interpretation -- or vague -- i.e., they do not clearly indicate where to draw the line when making a determination. Both ambiguity and vagueness can cause task indeterminacy -- the condition where some items in the evaluation corpus have more than one correct response. In this paper, we develop a framework for evaluating LLMs under task indeterminacy. Our framework disentangles the relationships between task specification, human ratings, and LLM responses in the LLM evaluation pipeline. Using our framework, we conduct a synthetic experiment showing that evaluations that use the "gold label" assumption underestimate the true performance. We also provide a method for estimating an error-adjusted performance interval given partial knowledge about indeterminate items in the evaluation corpus. We conclude by outlining implications of our work for the research community.

A Framework for Automated Measurement of Responsible AI Harms in Generative AI Applications

We present a framework for the automated measurement of responsible AI (RAI) metrics for large language models (LLMs) and associated products and services. Our framework for automatically measuring harms from LLMs builds on existing technical and sociotechnical expertise and leverages the capabilities of state-of-the-art LLMs, such as GPT-4. We use this framework to run through several case studies investigating how different LLMs may violate a range of RAI-related principles. The framework may be employed alongside domain-specific sociotechnical expertise to create measurements for new harm areas in the future. By implementing this framework, we aim to enable more advanced harm measurement efforts and further the responsible use of LLMs.

"One-size-fits-all"? Observations and Expectations of NLG Systems Across Identity-Related Language Features

Fairness-related assumptions about what constitutes appropriate NLG system behaviors range from invariance, where systems are expected to respond identically to social groups, to adaptation, where responses should instead vary across them. We design and conduct five case studies, in which we perturb different types of identity-related language features (names, roles, locations, dialect, and style) in NLG system inputs to illuminate tensions around invariance and adaptation. We outline people's expectations of system behaviors, and surface potential caveats of these two contrasting yet commonly-held assumptions. We find that motivations for adaptation include social norms, cultural differences, feature-specific information, and accommodation; motivations for invariance include perspectives that favor prescriptivism, view adaptation as unnecessary or too difficult for NLG systems to do appropriately, and are wary of false assumptions. Our findings highlight open challenges around defining what constitutes fair NLG system behavior.

Measuring Representational Harms in Image Captioning

Previous work has largely considered the fairness of image captioning systems through the underspecified lens of "bias." In contrast, we present a set of techniques for measuring five types of representational harms, as well as the resulting measurements obtained for two of the most popular image captioning datasets using a state-of-the-art image captioning system. Our goal was not to audit this image captioning system, but rather to develop normatively grounded measurement techniques, in turn providing an opportunity to reflect on the many challenges involved. We propose multiple measurement techniques for each type of harm. We argue that by doing so, we are better able to capture the multi-faceted nature of each type of harm, in turn improving the (collective) validity of the resulting measurements. Throughout, we discuss the assumptions underlying our measurement approach and point out when they do not hold.

Understanding Machine Learning Practitioners' Data Documentation Perceptions, Needs, Challenges, and Desiderata

Data is central to the development and evaluation of machine learning (ML) models. However, the use of problematic or inappropriate datasets can result in harms when the resulting models are deployed. To encourage responsible AI practice through more deliberate reflection on datasets and transparency around the processes by which they are created, researchers and practitioners have begun to advocate for increased data documentation and have proposed several data documentation frameworks. However, there is little research on whether these data documentation frameworks meet the needs of ML practitioners, who both create and consume datasets. To address this gap, we set out to understand ML practitioners' data documentation perceptions, needs, challenges, and desiderata, with the goal of deriving design requirements that can inform future data documentation frameworks. We conducted a series of semi-structured interviews with 14 ML practitioners at a single large, international technology company. We had them answer a list of questions taken from datasheets for datasets (Gebru, 2021). Our findings show that current approaches to data documentation are largely ad hoc and myopic in nature. Participants expressed needs for data documentation frameworks to be adaptable to their contexts, integrated into their existing tools and workflows, and automated wherever possible. Despite the fact that data documentation frameworks are often motivated from the perspective of responsible AI, participants did not make the connection between the questions that they were asked to answer and their responsible AI implications. In addition, participants often had difficulties prioritizing the needs of dataset consumers and providing information that someone unfamiliar with their datasets might need to know. Based on these findings, we derive seven design requirements for future data documentation frameworks.

REAL ML: Recognizing, Exploring, and Articulating Limitations of Machine Learning Research

Transparency around limitations can improve the scientific rigor of research, help ensure appropriate interpretation of research findings, and make research claims more credible. Despite these benefits, the machine learning (ML) research community lacks well-developed norms around disclosing and discussing limitations. To address this gap, we conduct an iterative design process with 30 ML and ML-adjacent researchers to develop and test REAL ML, a set of guided activities to help ML researchers recognize, explore, and articulate the limitations of their research. Using a three-stage interview and survey study, we identify ML researchers' perceptions of limitations, as well as the challenges they face when recognizing, exploring, and articulating limitations. We develop REAL ML to address some of these practical challenges, and highlight additional cultural challenges that will require broader shifts in community norms to address. We hope our study and REAL ML help move the ML research community toward more active and appropriate engagement with limitations.

Assessing the Fairness of AI Systems: AI Practitioners' Processes, Challenges, and Needs for Support

Various tools and practices have been developed to support practitioners in identifying, assessing, and mitigating fairness-related harms caused by AI systems. However, prior research has highlighted gaps between the intended design of these tools and practices and their use within particular contexts, including gaps caused by the role that organizational factors play in shaping fairness work. In this paper, we investigate these gaps for one such practice: disaggregated evaluations of AI systems, intended to uncover performance disparities between demographic groups. By conducting semi-structured interviews and structured workshops with thirty-three AI practitioners from ten teams at three technology companies, we identify practitioners' processes, challenges, and needs for support when designing disaggregated evaluations. We find that practitioners face challenges when choosing performance metrics, identifying the most relevant direct stakeholders and demographic groups on which to focus, and collecting datasets with which to conduct disaggregated evaluations. More generally, we identify impacts on fairness work stemming from a lack of engagement with direct stakeholders, business imperatives that prioritize customers over marginalized groups, and the drive to deploy AI systems at scale.