Research Papers

The Responsible Development of Automated Student Feedback with Generative AI

Euan D Lindsay, Mike Zhang, Aditya Johri +1

Providing rich, constructive feedback to students is essential for supporting and enhancing their learning. Recent advancements in Generative Artificial Intelligence (AI), particularly with large language models (LLMs), present new opportunities to deliver scalable, repeatable, and instant feedback, effectively making abundant a resource that has historically been scarce and costly. From a technical perspective, this approach is now feasible due to breakthroughs in AI and Natural Language Processing (NLP). While the potential educational benefits are compelling, implementing these technologies also introduces a host of ethical considerations that must be thoughtfully addressed. One of the core advantages of AI systems is their ability to automate routine and mundane tasks, potentially freeing up human educators for more nuanced work. However, the ease of automation risks a ``tyranny of the majority'', where the diverse needs of minority or unique learners are overlooked, as they may be harder to systematize and less straightforward to accommodate. Ensuring inclusivity and equity in AI-generated feedback, therefore, becomes a critical aspect of responsible AI implementation in education. The process of developing machine learning models that produce valuable, personalized, and authentic feedback also requires significant input from human domain experts. Decisions around whose expertise is incorporated, how it is captured, and when it is applied have profound implications for the relevance and quality of the resulting feedback. Additionally, the maintenance and continuous refinement of these models are necessary to adapt feedback to evolving contextual, theoretical, and student-related factors. Without ongoing adaptation, feedback risks becoming obsolete or mismatched with the current needs of diverse student populations [...]

Since the seminal paper by Breiman in 2001, who pointed out a potential harm of prediction multiplicities from the view of explainable AI, global analysis of a collection of all good models, also known as a `Rashomon set,' has been attracted much attention for the last years. Since finding such a set of good models is a hard computational problem, there have been only a few algorithms for the problem so far, most of which are either approximate or incomplete. To overcome this difficulty, we study efficient enumeration of all good models for a subclass of interpretable models, called rule lists. Based on a state-of-the-art optimal rule list learner, CORELS, proposed by Angelino et al. in 2017, we present an efficient enumeration algorithm CorelsEnum for exactly computing a set of all good models using polynomial space in input size, given a dataset and a error tolerance from an optimal model. By experiments with the COMPAS dataset on recidivism prediction, our algorithm CorelsEnum successfully enumerated all of several tens of thousands of good rule lists of length at most $\ell = 3$ in around 1,000 seconds, while a state-of-the-art top-$K$ rule list learner based on Lawler's method combined with CORELS, proposed by Hara and Ishihata in 2018, found only 40 models until the timeout of 6,000 seconds. For global analysis, we conducted experiments for characterizing the Rashomon set, and observed large diversity of models in predictive multiplicity and fairness of models.

Scientists frequently prioritize learning from data rather than training the best possible model; however, research in machine learning often prioritizes the latter. Marginal contribution feature importance (MCI) was developed to break this trend by providing a useful framework for quantifying the relationships in data. In this work, we aim to improve upon the theoretical properties, performance, and runtime of MCI by introducing ultra-marginal feature importance (UMFI), which uses dependence removal techniques from the AI fairness literature as its foundation. We first propose axioms for feature importance methods that seek to explain the causal and associative relationships in data, and we prove that UMFI satisfies these axioms under basic assumptions. We then show on real and simulated data that UMFI performs better than MCI, especially in the presence of correlated interactions and unrelated features, while partially learning the structure of the causal graph and reducing the exponential runtime of MCI to super-linear.

In today's society, AI systems are increasingly used to make critical decisions such as credit scoring and patient triage. However, great convenience brought by AI systems comes with troubling prevalence of bias against underrepresented groups. Mitigating bias in AI systems to increase overall fairness has emerged as an important challenge. Existing studies on mitigating bias in AI systems focus on eliminating sensitive demographic information embedded in data. Given the temporal and contextual complexity of conceptualizing fairness, lossy treatment of demographic information may contribute to an unnecessary trade-off between accuracy and fairness, especially when demographic attributes and class labels are correlated. In this paper, we present an information-lossless de-biasing technique that targets the scarcity of data in the disadvantaged group. Unlike the existing work, we demonstrate, both theoretically and empirically, that oversampling underrepresented groups can not only mitigate algorithmic bias in AI systems that consistently predict a favorable outcome for a certain group, but improve overall accuracy by mitigating class imbalance within data that leads to a bias towards the majority class. We demonstrate the effectiveness of our technique on real datasets using a variety of fairness metrics.

Modern face recognition systems leverage datasets containing images of hundreds of thousands of specific individuals' faces to train deep convolutional neural networks to learn an embedding space that maps an arbitrary individual's face to a vector representation of their identity. The performance of a face recognition system in face verification (1:1) and face identification (1:N) tasks is directly related to the ability of an embedding space to discriminate between identities. Recently, there has been significant public scrutiny into the source and privacy implications of large-scale face recognition training datasets such as MS-Celeb-1M and MegaFace, as many people are uncomfortable with their face being used to train dual-use technologies that can enable mass surveillance. However, the impact of an individual's inclusion in training data on a derived system's ability to recognize them has not previously been studied. In this work, we audit ArcFace, a state-of-the-art, open source face recognition system, in a large-scale face identification experiment with more than one million distractor images. We find a Rank-1 face identification accuracy of 79.71% for individuals present in the model's training data and an accuracy of 75.73% for those not present. This modest difference in accuracy demonstrates that face recognition systems using deep learning work better for individuals they are trained on, which has serious privacy implications when one considers all major open source face recognition training datasets do not obtain informed consent from individuals during their collection.

AI Agent Behavioral Science

Lin Chen, Yunke Zhang, Jie Feng +13

Recent advances in large language models (LLMs) have enabled the development of AI agents that exhibit increasingly human-like behaviors, including planning, adaptation, and social dynamics across diverse, interactive, and open-ended scenarios. These behaviors are not solely the product of the internal architectures of the underlying models, but emerge from their integration into agentic systems operating within specific contexts, where environmental factors, social cues, and interaction feedbacks shape behavior over time. This evolution necessitates a new scientific perspective: AI Agent Behavioral Science. Rather than focusing only on internal mechanisms, this perspective emphasizes the systematic observation of behavior, design of interventions to test hypotheses, and theory-guided interpretation of how AI agents act, adapt, and interact over time. We systematize a growing body of research across individual agent, multi-agent, and human-agent interaction settings, and further demonstrate how this perspective informs responsible AI by treating fairness, safety, interpretability, accountability, and privacy as behavioral properties. By unifying recent findings and laying out future directions, we position AI Agent Behavioral Science as a necessary complement to traditional model-centric approaches, providing essential tools for understanding, evaluating, and governing the real-world behavior of increasingly autonomous AI systems.

YouTube Recommendations Reinforce Negative Emotions: Auditing Algorithmic Bias with Emotionally-Agentic Sock Puppets

Hussam Habib, Rishab Nithyanand

Personalized recommendation algorithms, like those on YouTube, significantly shape online content consumption. These systems aim to maximize engagement by learning users' preferences and aligning content accordingly but may unintentionally reinforce impulsive and emotional biases. Using a sock-puppet audit methodology, this study examines YouTube's capacity to recognize and reinforce emotional preferences. Simulated user accounts with assigned emotional preferences navigate the platform, selecting videos that align with their assigned preferences and recording subsequent recommendations. Our findings reveal reveal that YouTube amplifies negative emotions, such as anger and grievance, by increasing their prevalence and prominence in recommendations. This reinforcement intensifies over time and persists across contexts. Surprisingly, contextual recommendations often exceed personalized ones in reinforcing emotional alignment. These findings suggest the algorithm amplifies user biases, contributing to emotional filter bubbles and raising concerns about user well-being and societal impacts. The study emphasizes the need for balancing personalization with content diversity and user agency.

AI fairness seeks to improve the transparency and explainability of AI systems by ensuring that their outcomes genuinely reflect the best interests of users. Data augmentation, which involves generating synthetic data from existing datasets, has gained significant attention as a solution to data scarcity. In particular, diffusion models have become a powerful technique for generating synthetic data, especially in fields like computer vision. This paper explores the potential of diffusion models to generate synthetic tabular data to improve AI fairness. The Tabular Denoising Diffusion Probabilistic Model (Tab-DDPM), a diffusion model adaptable to any tabular dataset and capable of handling various feature types, was utilized with different amounts of generated data for data augmentation. Additionally, reweighting samples from AIF360 was employed to further enhance AI fairness. Five traditional machine learning models-Decision Tree (DT), Gaussian Naive Bayes (GNB), K-Nearest Neighbors (KNN), Logistic Regression (LR), and Random Forest (RF)-were used to validate the proposed approach. Experimental results demonstrate that the synthetic data generated by Tab-DDPM improves fairness in binary classification.

We present a method for quantitative, in-depth analyses of fairness issues in AI systems with an application to credit scoring. To this aim we use BRIO, a tool for the evaluation of AI systems with respect to social unfairness and, more in general, ethically undesirable behaviours. It features a model-agnostic bias detection module, presented in \cite{DBLP:conf/beware/CoragliaDGGPPQ23}, to which a full-fledged unfairness risk evaluation module is added. As a case study, we focus on the context of credit scoring, analysing the UCI German Credit Dataset \cite{misc_statlog_(german_credit_data)_144}. We apply the BRIO fairness metrics to several, socially sensitive attributes featured in the German Credit Dataset, quantifying fairness across various demographic segments, with the aim of identifying potential sources of bias and discrimination in a credit scoring model. We conclude by combining our results with a revenue analysis.

Model fairness (a.k.a., bias) has become one of the most critical problems in a wide range of AI applications. An unfair model in autonomous driving may cause a traffic accident if corner cases (e.g., extreme weather) cannot be fairly regarded; or it will incur healthcare disparities if the AI model misdiagnoses a certain group of people (e.g., brown and black skin). In recent years, there have been emerging research works on addressing unfairness, and they mainly focus on a single unfair attribute, like skin tone; however, real-world data commonly have multiple attributes, among which unfairness can exist in more than one attribute, called 'multi-dimensional fairness'. In this paper, we first reveal a strong correlation between the different unfair attributes, i.e., optimizing fairness on one attribute will lead to the collapse of others. Then, we propose a novel Multi-Dimension Fairness framework, namely Muffin, which includes an automatic tool to unite off-the-shelf models to improve the fairness on multiple attributes simultaneously. Case studies on dermatology datasets with two unfair attributes show that the existing approach can achieve 21.05% fairness improvement on the first attribute while it makes the second attribute unfair by 1.85%. On the other hand, the proposed Muffin can unite multiple models to achieve simultaneously 26.32% and 20.37% fairness improvement on both attributes; meanwhile, it obtains 5.58% accuracy gain.

Biases in large-scale image datasets are known to influence the performance of computer vision models as a function of geographic context. To investigate the limitations of standard Internet data collection methods in low- and middle-income countries, we analyze human-centric image geo-diversity on a massive scale using geotagged Flickr images associated with each nation in Africa. We report the quantity and content of available data with comparisons to population-matched nations in Europe as well as the distribution of data according to fine-grained intra-national wealth estimates. Temporal analyses are performed at two-year intervals to expose emerging data trends. Furthermore, we present findings for an ``othering'' phenomenon as evidenced by a substantial number of images from Africa being taken by non-local photographers. The results of our study suggest that further work is required to capture image data representative of African people and their environments and, ultimately, to improve the applicability of computer vision models in a global context.

Intersectionality in Conversational AI Safety: How Bayesian Multilevel Models Help Understand Diverse Perceptions of Safety

Christopher M. Homan, Greg Serapio-Garcia, Lora Aroyo +5

Conversational AI systems exhibit a level of human-like behavior that promises to have profound impacts on many aspects of daily life -- how people access information, create content, and seek social support. Yet these models have also shown a propensity for biases, offensive language, and conveying false information. Consequently, understanding and moderating safety risks in these models is a critical technical and social challenge. Perception of safety is intrinsically subjective, where many factors -- often intersecting -- could determine why one person may consider a conversation with a chatbot safe and another person could consider the same conversation unsafe. In this work, we focus on demographic factors that could influence such diverse perceptions. To this end, we contribute an analysis using Bayesian multilevel modeling to explore the connection between rater demographics and how raters report safety of conversational AI systems. We study a sample of 252 human raters stratified by gender, age group, race/ethnicity group, and locale. This rater pool provided safety labels for 1,340 human-chatbot conversations. Our results show that intersectional effects involving demographic characteristics such as race/ethnicity, gender, and age, as well as content characteristics, such as degree of harm, all play significant roles in determining the safety of conversational AI systems. For example, race/ethnicity and gender show strong intersectional effects, particularly among South Asian and East Asian women. We also find that conversational degree of harm impacts raters of all race/ethnicity groups, but that Indigenous and South Asian raters are particularly sensitive to this harm. Finally, we observe the effect of education is uniquely intersectional for Indigenous raters, highlighting the utility of multilevel frameworks for uncovering underrepresented social perspectives.