Research Papers

Rejected job applicants seldom receive explanations from employers. Techniques from Explainable AI (XAI) could provide explanations at scale. Although XAI researchers have developed many different types of explanations, we know little about the type of explanations job applicants want. We use a survey of recent job applicants to fill this gap. Our survey generates three main insights. First, the current norm of, at most, generic feedback frustrates applicants. Second, applicants feel the employer has an obligation to provide an explanation. Third, job applicants want to know why they were unsuccessful and how to improve.

Neural network architectures have been extensively employed in the fair representation learning setting, where the objective is to learn a new representation for a given vector which is independent of sensitive information. Various "representation debiasing" techniques have been proposed in the literature. However, as neural networks are inherently opaque, these methods are hard to comprehend, which limits their usefulness. We propose a new framework for fair representation learning which is centered around the learning of "correction vectors", which have the same dimensionality as the given data vectors. The corrections are then simply summed up to the original features, and can therefore be analyzed as an explicit penalty or bonus to each feature. We show experimentally that a fair representation learning problem constrained in such a way does not impact performance.

Background: When neural network emotion and sentiment classifiers are used in public health informatics studies, biases present in the classifiers could produce inadvertently misleading results. Objective: This study assesses the impact of bias on COVID-19 topics, and demonstrates an automatic algorithm for reducing bias when applied to COVID-19 social media texts. This could help public health informatics studies produce more timely results during crises, with a reduced risk of misleading results. Methods: Emotion and sentiment classifiers were applied to COVID-19 data before and after debiasing the classifiers using unsupervised contrastive clustering. Contrastive clustering approximates the degree to which tokens exhibit a causal versus correlational relationship with emotion or sentiment, by contrasting the tokens' relative salience to topics versus emotions or sentiments. Results: Contrastive clustering distinguishes correlation from causation for tokens with an F1 score of 0.753. Masking bias prone tokens from the classifier input decreases the classifier's overall F1 score by 0.02 (anger) and 0.033 (negative sentiment), but improves the F1 score for sentences annotated as bias prone by 0.155 (anger) and 0.103 (negative sentiment). Averaging across topics, debiasing reduces anger estimates by 14.4% and negative sentiment estimates by 8.0%. Conclusions: Contrastive clustering reduces algorithmic bias in emotion and sentiment classification for social media text pertaining to the COVID-19 pandemic. Public health informatics studies should account for bias, due to its prevalence across a range of topics. Further research is needed to improve bias reduction techniques and to explore the adverse impact of bias on public health informatics analyses.

The paper introduces concepts of fairness and explainability (XAI) in artificial intelligence, oriented to solve a sophisticated business problems. For fairness, the authors discuss the bias-inducing specifics, as well as relevant mitigation methods, concluding with a set of recipes for introducing fairness in data-driven organizations. Additionally, for XAI, the authors audit specific algorithms paired with demonstrational business use-cases, discuss a plethora of techniques of explanations quality quantification and provide an overview of future research avenues.

The 3rd edition of the Montreal AI Ethics Institute's The State of AI Ethics captures the most relevant developments in AI Ethics since October 2020. It aims to help anyone, from machine learning experts to human rights activists and policymakers, quickly digest and understand the field's ever-changing developments. Through research and article summaries, as well as expert commentary, this report distills the research and reporting surrounding various domains related to the ethics of AI, including: algorithmic injustice, discrimination, ethical AI, labor impacts, misinformation, privacy, risk and security, social media, and more. In addition, The State of AI Ethics includes exclusive content written by world-class AI Ethics experts from universities, research institutes, consulting firms, and governments. Unique to this report is "The Abuse and Misogynoir Playbook," written by Dr. Katlyn Tuner (Research Scientist, Space Enabled Research Group, MIT), Dr. Danielle Wood (Assistant Professor, Program in Media Arts and Sciences; Assistant Professor, Aeronautics and Astronautics; Lead, Space Enabled Research Group, MIT) and Dr. Catherine D'Ignazio (Assistant Professor, Urban Science and Planning; Director, Data + Feminism Lab, MIT). The piece (and accompanying infographic), is a deep-dive into the historical and systematic silencing, erasure, and revision of Black women's contributions to knowledge and scholarship in the United Stations, and globally. Exposing and countering this Playbook has become increasingly important following the firing of AI Ethics expert Dr. Timnit Gebru (and several of her supporters) at Google. This report should be used not only as a point of reference and insight on the latest thinking in the field of AI Ethics, but should also be used as a tool for introspection as we aim to foster a more nuanced conversation regarding the impacts of AI on the world.

To implement fair machine learning in a sustainable way, choosing the right fairness objective is key. Since fairness is a concept of justice which comes in various, sometimes conflicting definitions, this is not a trivial task though. The most appropriate fairness definition for an artificial intelligence (AI) system is a matter of ethical standards and legal requirements, and the right choice depends on the particular use case and its context. In this position paper, we propose to use a decision tree as means to explain and justify the implemented kind of fairness to the end users. Such a structure would first of all support AI practitioners in mapping ethical principles to fairness definitions for a concrete application and therefore make the selection a straightforward and transparent process. However, this approach would also help document the reasoning behind the decision making. Due to the general complexity of the topic of fairness in AI, we argue that specifying "fairness" for a given use case is the best way forward to maintain confidence in AI systems. In this case, this could be achieved by sharing the reasons and principles expressed during the decision making process with the broader audience.

Machine learning software is increasingly being used to make decisions that affect people's lives. But sometimes, the core part of this software (the learned model), behaves in a biased manner that gives undue advantages to a specific group of people (where those groups are determined by sex, race, etc.). This "algorithmic discrimination" in the AI software systems has become a matter of serious concern in the machine learning and software engineering community. There have been works done to find "algorithmic bias" or "ethical bias" in the software system. Once the bias is detected in the AI software system, the mitigation of bias is extremely important. In this work, we a)explain how ground-truth bias in training data affects machine learning model fairness and how to find that bias in AI software,b)propose a methodFairwaywhich combines pre-processing and in-processing approach to remove ethical bias from training data and trained model. Our results show that we can find bias and mitigate bias in a learned model, without much damaging the predictive performance of that model. We propose that (1) test-ing for bias and (2) bias mitigation should be a routine part of the machine learning software development life cycle. Fairway offers much support for these two purposes.

A Set of Distinct Facial Traits Learned by Machines Is Not Predictive of Appearance Bias in the Wild

Ryan Steed, Aylin Caliskan

Research in social psychology has shown that people's biased, subjective judgments about another's personality based solely on their appearance are not predictive of their actual personality traits. But researchers and companies often utilize computer vision models to predict similarly subjective personality attributes such as "employability." We seek to determine whether state-of-the-art, black box face processing technology can learn human-like appearance biases. With features extracted with FaceNet, a widely used face recognition framework, we train a transfer learning model on human subjects' first impressions of personality traits in other faces as measured by social psychologists. We find that features extracted with FaceNet can be used to predict human appearance bias scores for deliberately manipulated faces but not for randomly generated faces scored by humans. Additionally, in contrast to work with human biases in social psychology, the model does not find a significant signal correlating politicians' vote shares with perceived competence bias. With Local Interpretable Model-Agnostic Explanations (LIME), we provide several explanations for this discrepancy. Our results suggest that some signals of appearance bias documented in social psychology are not embedded by the machine learning techniques we investigate. We shed light on the ways in which appearance bias could be embedded in face processing technology and cast further doubt on the practice of predicting subjective traits based on appearances.

This work showcases a new approach for causal discovery by leveraging user experiments and recent advances in photo-realistic image editing, demonstrating a potential of identifying causal factors and understanding complex systems counterfactually. We introduce the beauty learning problem as an example, which has been discussed metaphysically for centuries and been proved exists, is quantifiable, and can be learned by deep models in our recent paper, where we utilize a natural image generator coupled with user studies to infer causal effects from facial semantics to beauty outcomes, the results of which also align with existing empirical studies. We expect the proposed framework for a broader application in causal inference.

AI reviewers can assign different judgments to manuscripts that report the same science in different wording, potentially rewarding rhetorical optimization over scientific improvement. We formulate Rhetorical Robustness as the joint requirement of stability across content-preserving rewrites and discrimination across papers. We introduce RobustReview, a controlled full-manuscript benchmark with 1,260 manuscript versions, and evaluate 30 reviewer configurations. The benchmark reveals false robustness, where low rewrite sensitivity coincides with score collapse across papers, and shows that human alignment and rhetorical robustness rank reviewers differently. Moreover, the evaluated content-focused prompting protocol does not consistently improve robustness across backbones. Motivated by these findings, we introduce SciCore, a dual-branch reviewer that averages a full-manuscript judgment with a judgment based on an extracted, structured science core. This design combines manuscript-level assessment with a content-normalized view intended to reduce rhetorical sensitivity. In our primary GPT-5.5 comparison, SciCore achieves a leading joint stability-discrimination profile among the benchmarked reviewers while maintaining competitive human alignment. These results identify rhetorical robustness as a distinct evaluation target and demonstrate the potential of science-core review to improve it.

Interpretability and fairness are two of the most emphasized dimensions in trustworthy artificial intelligence (AI). Various explainable AI methods have been introduced to improve interpretability. This paper focuses on neural network (NN)-based generalized additive models (GAMs), a class of self-interpretable models. While most existing research has prioritized improving the accuracy of NN-based GAMs, their interpretability remains largely underexplored. To address this gap, this paper introduces explicit quantitative metrics for evaluating the interpretability of NN-based GAMs, empirically examines their effectiveness, and explores strategies for improving interpretability within these models. In addition, the simultaneous and explicit optimization of both interpretability and fairness, along with their trade-offs and the underlying reasons, remains underexplored. To address this, we propose a multi-objective neural basis model (MONBM) framework based on multi-objective evolutionary learning to consider accuracy, interpretability, and fairness simultaneously. A partial retraining strategy is further developed to facilitate the practical application of evolutionary multi-objective optimization to deep model architectures. Based on MONBM, this paper reveals the complex relationships between these dimensions and the reasons behind these intricate relationships. This analysis demonstrates how multi-objective optimization can be combined with self-interpretable models to reveal relationships among trustworthiness objectives. In addition, MONBM obtains a set of models with different trade-offs between dimensions, and the competitiveness of the approach is validated by comparing it with state-of-the-art methods.

Stress-Testing Efficient Responsible-AI Evaluation: When Compute Savings Change Benchmark Conclusions

Ahmed El Kady, Aravind Narayanan, Rehana Riaz +2

Efficient evaluation changes the protocol used to support claims about model behavior, yet it is rarely tested whether those claims remain stable after the evaluation itself is made cheaper. We stress-test conclusion robustness in responsible-AI benchmarking by evaluating three dense and mixture-of-experts models on BBQ and BBQ-V under seven conditions spanning batching, quantization, benchmark reduction, and their combinations. Rather than treating preserved aggregate accuracy as sufficient, we compare accuracy, bias severity and prevalence, reasoning quality, subgroup behavior, subset-membership stability, runtime, and measured GPU energy against a full-benchmark BF16 baseline. Larger batching keeps accuracy within 0.35 percentage points of baseline and produces comparatively small subgroup changes, while reducing energy in five of six model--dataset settings. INT8 largely preserves quality but uses 1.79--4.26$\times$ baseline energy. INT4 causes larger, model- and context-dependent changes. Reduced benchmarks provide the most consistent savings, but very small subsets are substantially more sensitive to which items are retained. Efficient evaluation should therefore be treated as a measurement intervention whose validity must be checked across the conclusions the benchmark is intended to support. Our project website is https://vectorinstitute.github.io/sustainable-rai-evaluation/ and the code is available at https://github.com/VectorInstitute/sustainable-rai-evaluation.