Search

Large language models (LLMs) have been shown to be effective on tabular prediction tasks in the low-data regime, leveraging their internal knowledge and ability to learn from instructions and examples. However, LLMs can fail to generate predictions that satisfy group fairness, that is, produce equitable outcomes across groups. Critically, conventional debiasing approaches for natural language tasks do not directly translate to mitigating group unfairness in tabular settings. In this work, we systematically investigate four empirical approaches to improve group fairness of LLM predictions on tabular datasets, including fair prompt optimization, soft prompt tuning, strategic selection of few-shot examples, and self-refining predictions via chain-of-thought reasoning. Through experiments on four tabular datasets using both open-source and proprietary LLMs, we show the effectiveness of these methods in enhancing demographic parity while maintaining high overall performance. Our analysis provides actionable insights for practitioners in selecting the most suitable approach based on their specific requirements and constraints.

How can we precisely estimate a large language model's (LLM) accuracy on questions belonging to a specific topic within a larger question-answering dataset? The standard direct estimator, which averages the model's accuracy on the questions in each subgroup, may exhibit high variance for subgroups (topics) with small sample sizes. Synthetic regression modeling, which leverages the model's accuracy on questions about other topics, may yield biased estimates that are too unreliable for large subgroups. We prescribe a simple yet effective solution: an empirical Bayes (EB) estimator that balances direct and regression estimates for each subgroup separately, improving the precision of subgroup-level estimates of model performance. Our experiments on multiple datasets show that this approach consistently provides more precise estimates of the LLM performance compared to the direct and regression approaches, achieving substantial reductions in the mean squared error. Confidence intervals for EB estimates also have near-nominal coverage and are narrower compared to those for the direct estimator. Additional experiments on tabular and vision data validate the benefits of this EB approach., Comment: To appear at EMNLP 2024

As recommender systems become widely deployed in different domains, they increasingly influence their users' beliefs and preferences. Auditing recommender systems is crucial as it not only ensures the continuous improvement of recommendation algorithms but also safeguards against potential issues like biases and ethical concerns. In this paper, we view recommender system auditing from a causal lens and provide a general recipe for defining auditing metrics. Under this general causal auditing framework, we categorize existing auditing metrics and identify gaps in them -- notably, the lack of metrics for auditing user agency while accounting for the multi-step dynamics of the recommendation process. We leverage our framework and propose two classes of such metrics:future- and past-reacheability and stability, that measure the ability of a user to influence their own and other users' recommendations, respectively. We provide both a gradient-based and a black-box approach for computing these metrics, allowing the auditor to compute them under different levels of access to the recommender system. In our experiments, we demonstrate the efficacy of methods for computing the proposed metrics and inspect the design of recommender systems through these proposed metrics., Comment: 28 pages

Recent philosophical work has explored how the social identity of knowers influences how their contributions are received, assessed, and credited. However, a critical gap remains regarding the role of technology in mediating and enabling communication within today's epistemic communities. This paper addresses this gap by examining how social media platforms and their recommendation algorithms shape the professional visibility and opportunities of researchers from minority groups. Using agent-based simulations, we investigate this question with respect to components of a widely used recommendation algorithm, and uncover three key patterns: First, these algorithms disproportionately harm the professional visibility of researchers from minority groups, creating systemic patterns of exclusion. Second, within these minority groups, the algorithms result in greater visibility for users who more closely resemble the majority group, incentivizing assimilation at the cost of professional invisibility. Third, even for topics that strongly align with minority identities, content created by minority researchers is less visible to the majority than similar content produced by majority users. Importantly, these patterns emerge, even though individual engagement with professional content is independent of group identity. These findings have significant implications for philosophical discussions on epistemic injustice and exclusion, and for policy proposals aimed at addressing these harms. More broadly, they call for a closer examination of the pervasive, but often neglected role of AI and data-driven technologies in shaping today's epistemic communities.

Identifying the causal pathways of unfairness is a critical objective for improving policy design and algorithmic decision-making. Prior work in causal fairness analysis often requires knowledge of the causal graph, hindering practical applications in complex or low-knowledge domains. Moreover, global discovery methods that learn causal structure from data can display unstable performance on finite samples, preventing robust fairness conclusions. To mitigate these challenges, we introduce local discovery for direct discrimination (LD3): a method that uncovers structural evidence of direct unfairness by identifying the causal parents of an outcome variable. LD3 performs a linear number of conditional independence tests relative to variable set size, and allows for latent confounding under the sufficient condition that all parents of the outcome are observed. We show that LD3 returns a valid adjustment set (VAS) under a new graphical criterion for the weighted controlled direct effect, a qualitative indicator of direct discrimination. LD3 limits unnecessary adjustment, providing interpretable VAS for assessing unfairness. We use LD3 to analyze causal fairness in two complex decision systems: criminal recidivism prediction and liver transplant allocation. LD3 was more time-efficient and returned more plausible results on real-world data than baselines, which took 46$\times$ to 5870$\times$ longer to execute.

Predictive risk models in the public sector are commonly developed using administrative data that is more complete for subpopulations that more greatly rely on public services. In the United States, for instance, information on health care utilization is routinely available to government agencies for individuals supported by Medicaid and Medicare, but not for the privately insured. Critiques of public sector algorithms have identified such differential feature under-reporting as a driver of disparities in algorithmic decision-making. Yet this form of data bias remains understudied from a technical viewpoint. While prior work has examined the fairness impacts of additive feature noise and features that are clearly marked as missing, the setting of data missingness absent indicators (i.e. differential feature under-reporting) has been lacking in research attention. In this work, we present an analytically tractable model of differential feature under-reporting which we then use to characterize the impact of this kind of data bias on algorithmic fairness. We demonstrate how standard missing data methods typically fail to mitigate bias in this setting, and propose a new set of methods specifically tailored to differential feature under-reporting. Our results show that, in real world data settings, under-reporting typically leads to increasing disparities. The proposed solution methods show success in mitigating increases in unfairness., Comment: ACM Conference on Fairness, Accountability, and Transparency (FAccT 2024)

Motivated by the growing importance of reducing unfairness in ML predictions, Fair-ML researchers have presented an extensive suite of algorithmic 'fairness-enhancing' remedies. Most existing algorithms, however, are agnostic to the sources of the observed unfairness. As a result, the literature currently lacks guiding frameworks to specify conditions under which each algorithmic intervention can potentially alleviate the underpinning cause of unfairness. To close this gap, we scrutinize the underlying biases (e.g., in the training data or design choices) that cause observational unfairness. We present the conceptual idea and a first implementation of a bias-injection sandbox tool to investigate fairness consequences of various biases and assess the effectiveness of algorithmic remedies in the presence of specific types of bias. We call this process the bias(stress)-testing of algorithmic interventions. Unlike existing toolkits, ours provides a controlled environment to counterfactually inject biases in the ML pipeline. This stylized setup offers the distinct capability of testing fairness interventions beyond observational data and against an unbiased benchmark. In particular, we can test whether a given remedy can alleviate the injected bias by comparing the predictions resulting after the intervention in the biased setting with true labels in the unbiased regime-that is, before any bias injection. We illustrate the utility of our toolkit via a proof-of-concept case study on synthetic data. Our empirical analysis showcases the type of insights that can be obtained through our simulations., Comment: Appeared as a poster at the second ACM conference on Equity and Access in Algorithms, Mechanisms, and Optimization (EAAMO'22)

Recommender system fairness has been studied from the perspectives of a variety of stakeholders including content producers, the content itself and recipients of recommendations. Regardless of which type of stakeholders are considered, most works in this area assess the efficacy of fairness intervention by evaluating a single fixed fairness criterion through the lens of a one-shot, static setting. Yet recommender systems constitute dynamical systems with feedback loops from the recommendations to the underlying population distributions which could lead to unforeseen and adverse consequences if not taken into account. In this paper, we study a connection recommender system patterned after the systems employed by web-scale social networks and analyze the long-term effects of intervening on fairness in the recommendations. We find that, although seemingly fair in aggregate, common exposure and utility parity interventions fail to mitigate amplification of biases in the long term. We theoretically characterize how certain fairness interventions impact the bias amplification dynamics in a stylized P\'{o}lya urn model., Comment: Conference on Artificial Intelligence, Ethics, and Society (AIES 2022)

Police departments around the world have been experimenting with forms of place-based data-driven proactive policing for over two decades. Modern incarnations of such systems are commonly known as hot spot predictive policing. These systems predict where future crime is likely to concentrate such that police can allocate patrols to these areas and deter crime before it occurs. Previous research on fairness in predictive policing has concentrated on the feedback loops which occur when models are trained on discovered crime data, but has limited implications for models trained on victim crime reporting data. We demonstrate how differential victim crime reporting rates across geographical areas can lead to outcome disparities in common crime hot spot prediction models. Our analysis is based on a simulation patterned after district-level victimization and crime reporting survey data for Bogot\'a, Colombia. Our results suggest that differential crime reporting rates can lead to a displacement of predicted hotspots from high crime but low reporting areas to high or medium crime and high reporting areas. This may lead to misallocations both in the form of over-policing and under-policing., Comment: Conference on Fairness, Accountability, and Transparency (FAccT 2021)

Learning to predict solutions to real-valued combinatorial graph problems promises efficient approximations. As demonstrated based on the NP-hard edge clique cover number, recurrent neural networks (RNNs) are particularly suited for this task and can even outperform state-of-the-art heuristics. However, the theoretical framework for estimating real-valued RNNs is understood only poorly. As our primary contribution, this is the first work that upper bounds the sample complexity for learning real-valued RNNs. While such derivations have been made earlier for feed-forward and convolutional neural networks, our work presents the first such attempt for recurrent neural networks. Given a single-layer RNN with $a$ rectified linear units and input of length $b$, we show that a population prediction error of $\varepsilon$ can be realized with at most $\tilde{\mathcal{O}}(a^4b/\varepsilon^2)$ samples. We further derive comparable results for multi-layer RNNs. Accordingly, a size-adaptive RNN fed with graphs of at most $n$ vertices can be learned in $\tilde{\mathcal{O}}(n^6/\varepsilon^2)$, i.e., with only a polynomial number of samples. For combinatorial graph problems, this provides a theoretical foundation that renders RNNs competitive., Comment: A two-page summary of this paper has been accepted as a student abstract at AAAI-20, this is the extended full version