Kristian Lum

Kristian Lum, Yunfeng Zhang, Amanda Bower

When a model's performance differs across socially or culturally relevant groups--like race, gender, or the intersections of many such groups--it is often called "biased." While much of the work in algorithmic fairness over the last several years has focused on developing various definitions of model fairness (the absence of group-wise model performance disparities) and eliminating such "bias," much less work has gone into rigorously measuring it. In practice, it important to have high quality, human digestible measures of model performance disparities and associated uncertainty quantification about them that can serve as inputs into multi-faceted decision-making processes. In this paper, we show both mathematically and through simulation that many of the metrics used to measure group-wise model performance disparities are themselves statistically biased estimators of the underlying quantities they purport to represent. We argue that this can cause misleading conclusions about the relative group-wise model performance disparities along different dimensions, especially in cases where some sensitive variables consist of categories with few members. We propose the "double-corrected" variance estimator, which provides unbiased estimates and uncertainty quantification of the variance of model performance across groups. It is conceptually simple and easily implementable without statistical software package or numerical optimization. We demonstrate the utility of this approach through simulation and show on a real dataset that while statistically biased estimators of group-wise model performance disparities indicate statistically significant differences, when accounting for statistical bias in the estimator, the estimated between-group disparities are no longer statistically significant.

Mohsen Abbasi, Suresh Venkatasubramanian, Sorelle A. Friedler et al.

Voter suppression and associated racial disparities in access to voting are long-standing civil rights concerns in the United States. Barriers to voting have taken many forms over the decades. A history of violent explicit discouragement has shifted to more subtle access limitations that can include long lines and wait times, long travel times to reach a polling station, and other logistical barriers to voting. Our focus in this work is on quantifying disparities in voting access pertaining to the overall time-to-vote, and how they could be remedied via a better choice of polling location or provisioning more sites where voters can cast ballots. However, appropriately calibrating access disparities is difficult because of the need to account for factors such as population density and different community expectations for reasonable travel times. In this paper, we quantify access to polling locations, developing a methodology for the calibrated measurement of racial disparities in polling location "load" and distance to polling locations. We apply this methodology to a study of real-world data from Florida and North Carolina to identify disparities in voting access from the 2020 election. We also introduce algorithms, with modifications to handle scale, that can reduce these disparities by suggesting new polling locations from a given list of identified public locations (including schools and libraries). Applying these algorithms on the 2020 election location data also helps to expose and explore tradeoffs between the cost of allocating more polling locations and the potential impact on access disparities. The developed voting access measurement methodology and algorithmic remediation technique is a first step in better polling location assignment.

Canfer Akbulut, Rasmi Elasmar, Abhishek Roy et al.

Interest in the concept of AI-driven harmful manipulation is growing, yet current approaches to evaluating it are limited. This paper introduces a framework for evaluating harmful AI manipulation via context-specific human-AI interaction studies. We illustrate the utility of this framework by assessing an AI model with 10,101 participants spanning interactions in three AI use domains (public policy, finance, and health) and three locales (US, UK, and India). Overall, we find that that the tested model can produce manipulative behaviours when prompted to do so and, in experimental settings, is able to induce belief and behaviour changes in study participants. We further find that context matters: AI manipulation differs between domains, suggesting that it needs to be evaluated in the high-stakes context(s) in which an AI system is likely to be used. We also identify significant differences across our tested geographies, suggesting that AI manipulation results from one geographic region may not generalise to others. Finally, we find that the frequency of manipulative behaviours (propensity) of an AI model is not consistently predictive of the likelihood of manipulative success (efficacy), underscoring the importance of studying these dimensions separately. To facilitate adoption of our evaluation framework, we detail our testing protocols and make relevant materials publicly available. We conclude by discussing open challenges in evaluating harmful manipulation by AI models.

Imagen-Team-Google, Jason Baldridge, Jakob Bauer et al.

We introduce Imagen 3, a latent diffusion model that generates high quality images from text prompts. We describe our quality and responsibility evaluations. Imagen 3 is preferred over other state-of-the-art (SOTA) models at the time of evaluation. In addition, we discuss issues around safety and representation, as well as methods we used to minimize the potential harm of our models.

Kristian Lum, Jacy Reese Anthis, Kevin Robinson et al.

Standard benchmarks of bias and fairness in large language models (LLMs) measure the association between the user attributes stated or implied by a prompt and the LLM's short text response, but human-AI interaction increasingly requires long-form and context-specific system output to solve real-world tasks. In the commonly studied domain of gender-occupation bias, we test whether these benchmarks are robust to lengthening the LLM responses as a measure of Realistic Use and Tangible Effects (i.e., RUTEd evaluations). From the current literature, we adapt three standard bias metrics (neutrality, skew, and stereotype) and develop analogous RUTEd evaluations from three contexts of real-world use: children's bedtime stories, user personas, and English language learning exercises. We find that standard bias metrics have no significant correlation with the more realistic bias metrics. For example, selecting the least biased model based on the standard "trick tests" coincides with selecting the least biased model as measured in more realistic use no more than random chance. We suggest that there is not yet evidence to justify standard benchmarks as reliable proxies of real-world AI biases, and we encourage further development of evaluations grounded in particular contexts.

Johannes Himmelreich, Arbie Hsu, Kristian Lum et al.

A yet unmet challenge in algorithmic fairness is the problem of intersectionality, that is, achieving fairness across the intersection of multiple groups -- and verifying that such fairness has been attained. Because intersectional groups tend to be small, verifying whether a model is fair raises statistical as well as moral-methodological challenges. This paper (1) elucidates the problem of intersectionality in algorithmic fairness, (2) develops desiderata to clarify the challenges underlying the problem and guide the search for potential solutions, (3) illustrates the desiderata and potential solutions by sketching a proposal using simple hypothesis testing, and (4) evaluates, partly empirically, this proposal against the proposed desiderata.

Laura Weidinger, John Mellor, Bernat Guillen Pegueroles et al.

This research introduces STAR, a sociotechnical framework that improves on current best practices for red teaming safety of large language models. STAR makes two key contributions: it enhances steerability by generating parameterised instructions for human red teamers, leading to improved coverage of the risk surface. Parameterised instructions also provide more detailed insights into model failures at no increased cost. Second, STAR improves signal quality by matching demographics to assess harms for specific groups, resulting in more sensitive annotations. STAR further employs a novel step of arbitration to leverage diverse viewpoints and improve label reliability, treating disagreement not as noise but as a valuable contribution to signal quality.

Tomo Lazovich, Luca Belli, Aaron Gonzales et al.

The harmful impacts of algorithmic decision systems have recently come into focus, with many examples of systems such as machine learning (ML) models amplifying existing societal biases. Most metrics attempting to quantify disparities resulting from ML algorithms focus on differences between groups, dividing users based on demographic identities and comparing model performance or overall outcomes between these groups. However, in industry settings, such information is often not available, and inferring these characteristics carries its own risks and biases. Moreover, typical metrics that focus on a single classifier's output ignore the complex network of systems that produce outcomes in real-world settings. In this paper, we evaluate a set of metrics originating from economics, distributional inequality metrics, and their ability to measure disparities in content exposure in a production recommendation system, the Twitter algorithmic timeline. We define desirable criteria for metrics to be used in an operational setting, specifically by ML practitioners. We characterize different types of engagement with content on Twitter using these metrics, and use these results to evaluate the metrics with respect to the desired criteria. We show that we can use these metrics to identify content suggestion algorithms that contribute more strongly to skewed outcomes between users. Overall, we conclude that these metrics can be useful tools for understanding disparate outcomes in online social networks.

Kristian Lum, James Johndrow

Predictive modeling is increasingly being employed to assist human decision-makers. One purported advantage of replacing human judgment with computer models in high stakes settings-- such as sentencing, hiring, policing, college admissions, and parole decisions-- is the perceived "neutrality" of computers. It is argued that because computer models do not hold personal prejudice, the predictions they produce will be equally free from prejudice. There is growing recognition that employing algorithms does not remove the potential for bias, and can even amplify it, since training data were inevitably generated by a process that is itself biased. In this paper, we provide a probabilistic definition of algorithmic bias. We propose a method to remove bias from predictive models by removing all information regarding protected variables from the permitted training data. Unlike previous work in this area, our framework is general enough to accommodate arbitrary data types, e.g. binary, continuous, etc. Motivated by models currently in use in the criminal justice system that inform decisions on pre-trial release and paroling, we apply our proposed method to a dataset on the criminal histories of individuals at the time of sentencing to produce "race-neutral" predictions of re-arrest. In the process, we demonstrate that the most common approach to creating "race-neutral" models-- omitting race as a covariate-- still results in racially disparate predictions. We then demonstrate that the application of our proposed method to these data removes racial disparities from predictions with minimal impact on predictive accuracy.