Research Papers

During job recruitment, traditional applicant selection methods often lack transparency. Candidates are rarely given sufficient justifications for recruiting decisions, whether they are made manually by human recruiters or through the use of black-box Applicant Tracking Systems (ATS). To address this problem, our work introduces a multi-agent AI system that uses Large Language Models (LLMs) to guide job seekers during the recruitment process. Using an iterative user-centric design approach, we first conducted a two-phased exploratory study with four active job seekers to inform the design and development of the system. Subsequently, we conducted an in-depth, qualitative user study with 20 active job seekers through individual one-to-one interviews to evaluate the developed prototype. The results of our evaluation demonstrate that participants perceived our multi-agent recruitment system as significantly more actionable, trustworthy, and fair compared to traditional methods. Our study further helped us uncover in-depth insights into factors contributing to these perceived user experiences. Drawing from these insights, we offer broader design implications for building user-aligned, multi-agent explainable AI systems across diverse domains.

As Large Language Models (LLMs) become increasingly integrated into real-world decision-making systems, understanding their behavioural vulnerabilities remains a critical challenge for AI safety and alignment. While existing evaluation metrics focus primarily on reasoning accuracy or factual correctness, they often overlook whether LLMs are robust to adversarial manipulation or capable of using adaptive strategy in dynamic environments. This paper introduces an adversarial evaluation framework designed to systematically stress-test the decision-making processes of LLMs under interactive and adversarial conditions. Drawing on methodologies from cognitive psychology and game theory, our framework probes how models respond in two canonical tasks: the two-armed bandit task and the Multi-Round Trust Task. These tasks capture key aspects of exploration-exploitation trade-offs, social cooperation, and strategic flexibility. We apply this framework to several state-of-the-art LLMs, including GPT-3.5, GPT-4, Gemini-1.5, and DeepSeek-V3, revealing model-specific susceptibilities to manipulation and rigidity in strategy adaptation. Our findings highlight distinct behavioral patterns across models and emphasize the importance of adaptability and fairness recognition for trustworthy AI deployment. Rather than offering a performance benchmark, this work proposes a methodology for diagnosing decision-making weaknesses in LLM-based agents, providing actionable insights for alignment and safety research.

Trustworthy AI encompasses many aspirational aspects for aligning AI systems with human values, including fairness, privacy, robustness, explainability, and uncertainty quantification. The ultimate goal of Trustworthy AI research is to achieve all aspects simultaneously. However, efforts to enhance one aspect often introduce unintended trade-offs that negatively impact others. In this position paper, we review notable approaches to five aspects and systematically consider every pair, detailing the negative interactions that can arise. For example, applying differential privacy to model training can amplify biases, undermining fairness. Drawing on these findings, we take the position that current research practices of improving one or two aspects in isolation are insufficient. Instead, research on Trustworthy AI must account for interactions between aspects and adopt a holistic view across all relevant axes at once. We offer recommendations for how researchers can work towards integrated trust, and provide guidance for practitioners to manage interactions today.

Given the demand for responsible and trustworthy AI for education, this study evaluates symbolic, sub-symbolic, and neural-symbolic AI (NSAI) in terms of generalizability and interpretability. Our extensive experiments on balanced and imbalanced self-regulated learning datasets of Estonian primary school students predicting 7th-grade mathematics national test performance showed that symbolic and sub-symbolic methods performed well on balanced data but struggled to identify low performers in imbalanced datasets. Interestingly, symbolic and sub-symbolic methods emphasized different factors in their decision-making: symbolic approaches primarily relied on cognitive and motivational factors, while sub-symbolic methods focused more on cognitive aspects, learnt knowledge, and the demographic variable of gender -- yet both largely overlooked metacognitive factors. The NSAI method, on the other hand, showed advantages by: (i) being more generalizable across both classes -- even in imbalanced datasets -- as its symbolic knowledge component compensated for the underrepresented class; and (ii) relying on a more integrated set of factors in its decision-making, including motivation, (meta)cognition, and learnt knowledge, thus offering a comprehensive and theoretically grounded interpretability framework. These contrasting findings highlight the need for a holistic comparison of AI methods before drawing conclusions based solely on predictive performance. They also underscore the potential of hybrid, human-centred NSAI methods to address the limitations of other AI families and move us closer to responsible AI for education. Specifically, by enabling stakeholders to contribute to AI design, NSAI aligns learned patterns with theoretical constructs, incorporates factors like motivation and metacognition, and strengthens the trustworthiness and responsibility of educational data mining.

Machine learning (ML) models may suffer from significant performance disparities between patient groups. Identifying such disparities by monitoring performance at a granular level is crucial for safely deploying ML to each patient. Traditional subgroup analysis based on metadata can expose performance disparities only if the available metadata (e.g., patient sex) sufficiently reflects the main reasons for performance variability, which is not common. Subgroup discovery techniques that identify cohesive subgroups based on learned feature representations appear as a potential solution: They could expose hidden stratifications and provide more granular subgroup performance reports. However, subgroup discovery is challenging to evaluate even as a standalone task, as ground truth stratification labels do not exist in real data. Subgroup discovery has thus neither been applied nor evaluated for the application of subgroup performance monitoring. Here, we apply subgroup discovery for performance monitoring in chest x-ray and skin lesion classification. We propose novel evaluation strategies and show that a simplified subgroup discovery method without access to classification labels or metadata can expose larger performance disparities than traditional metadata-based subgroup analysis. We provide the first compelling evidence that subgroup discovery can serve as an important tool for comprehensive performance validation and monitoring of trustworthy AI in medicine.

Rethinking Prompt-based Debiasing in Large Language Models

Xinyi Yang, Runzhe Zhan, Derek F. Wong +3

Investigating bias in large language models (LLMs) is crucial for developing trustworthy AI. While prompt-based through prompt engineering is common, its effectiveness relies on the assumption that models inherently understand biases. Our study systematically analyzed this assumption using the BBQ and StereoSet benchmarks on both open-source models as well as commercial GPT model. Experimental results indicate that prompt-based is often superficial; for instance, the Llama2-7B-Chat model misclassified over 90% of unbiased content as biased, despite achieving high accuracy in identifying bias issues on the BBQ dataset. Additionally, specific evaluation and question settings in bias benchmarks often lead LLMs to choose "evasive answers", disregarding the core of the question and the relevance of the response to the context. Moreover, the apparent success of previous methods may stem from flawed evaluation metrics. Our research highlights a potential "false prosperity" in prompt-base efforts and emphasizes the need to rethink bias metrics to ensure truly trustworthy AI.

The growth of Educational Technology (EdTech) has enabled highly personalized learning experiences through Artificial Intelligence (AI)-based recommendation systems tailored to each student needs. However, these systems can unintentionally introduce biases, potentially limiting fair access to learning resources. This study presents a recommendation system for K-12 students, combining graph-based modeling and matrix factorization to provide personalized suggestions for extracurricular activities, learning resources, and volunteering opportunities. To address fairness concerns, the system includes a framework to detect and reduce biases by analyzing feedback across protected student groups. This work highlights the need for continuous monitoring in educational recommendation systems to support equitable, transparent, and effective learning opportunities for all students.

We fine-tuned and compared several encoder-based Transformer large language models (LLM) to predict differential item functioning (DIF) from the item text. We then applied explainable artificial intelligence (XAI) methods to these models to identify specific words associated with DIF. The data included 42,180 items designed for English language arts and mathematics summative state assessments among students in grades 3 to 11. Prediction $R^2$ ranged from .04 to .32 among eight focal and reference group pairs. Our findings suggest that many words associated with DIF reflect minor sub-domains included in the test blueprint by design, rather than construct-irrelevant item content that should be removed from assessments. This may explain why qualitative reviews of DIF items often yield confusing or inconclusive results. Our approach can be used to screen words associated with DIF during the item-writing process for immediate revision, or help review traditional DIF analysis results by highlighting key words in the text. Extensions of this research can enhance the fairness of assessment programs, especially those that lack resources to build high-quality items, and among smaller subpopulations where we do not have sufficient sample sizes for traditional DIF analyses.

This paper presents a survey of local US policymakers' views on the future impact and regulation of AI. Our survey provides insight into US policymakers' expectations regarding the effects of AI on local communities and the nation, as well as their attitudes towards specific regulatory policies. Conducted in two waves (2022 and 2023), the survey captures changes in attitudes following the release of ChatGPT and the subsequent surge in public awareness of AI. Local policymakers express a mix of concern, optimism, and uncertainty about AI's impacts, anticipating significant societal risks such as increased surveillance, misinformation, and political polarization, alongside potential benefits in innovation and infrastructure. Many also report feeling underprepared and inadequately informed to make AI-related decisions. On regulation, a majority of policymakers support government oversight and favor specific policies addressing issues such as data privacy, AI-related unemployment, and AI safety and fairness. Democrats show stronger and more consistent support for regulation than Republicans, but the latter experienced a notable shift towards majority support between 2022 and 2023. Our study contributes to understanding the perspectives of local policymakers-key players in shaping state and federal AI legislation-by capturing evolving attitudes, partisan dynamics, and their implications for policy formation. The findings highlight the need for capacity-building initiatives and bi-partisan coordination to mitigate policy fragmentation and build a cohesive framework for AI governance in the US.

Artificial intelligence (AI) and machine learning (ML) have made tremendous advancements in the past decades. From simple recommendation systems to more complex tumor identification systems, AI/ML systems have been utilized in a plethora of applications. This rapid growth of AI/ML and its proliferation in numerous private and public sector applications, while successful, has also opened new challenges and obstacles for regulators. With almost little to no human involvement required for some of the new decision-making AI/ML systems, there is now a pressing need to ensure the responsible use of these systems. Particularly in federal government use-cases, the use of AI technologies must be carefully governed by appropriate transparency and accountability mechanisms. This has given rise to new interdisciplinary fields of AI research such as \textit{Responsible AI (RAI)}. In this position paper we provide a brief overview of development in RAI and discuss some of the motivating principles commonly explored in the field. An overview of the current regulatory landscape relating to AI is also discussed with analysis of different Executive Orders, policies and frameworks. We then present examples of how federal agencies are aiming for the responsible use of AI, specifically we present use-case examples of different projects and research from the Census Bureau on implementing the responsible use of AI. We also provide a brief overview for a Responsible AI Assessment Toolkit currently under-development aimed at helping federal agencies operationalize RAI principles. Finally, a robust discussion on how different policies/regulations map to RAI principles, along with challenges and opportunities for regulation/governance of responsible AI within the federal government is presented.

Malaria remains a significant global health burden, particularly in resource-limited regions where timely and accurate diagnosis is critical to effective treatment and control. Deep Learning (DL) has emerged as a transformative tool for automating malaria detection and it offers high accuracy and scalability. However, the effectiveness of these models is constrained by challenges in data quality and model generalization including imbalanced datasets, limited diversity and annotation variability. These issues reduce diagnostic reliability and hinder real-world applicability. This article provides a comprehensive analysis of these challenges and their implications for malaria detection performance. Key findings highlight the impact of data imbalances which can lead to a 20\% drop in F1-score and regional biases which significantly hinder model generalization. Proposed solutions, such as GAN-based augmentation, improved accuracy by 15-20\% by generating synthetic data to balance classes and enhance dataset diversity. Domain adaptation techniques, including transfer learning, further improved cross-domain robustness by up to 25\% in sensitivity. Additionally, the development of diverse global datasets and collaborative data-sharing frameworks is emphasized as a cornerstone for equitable and reliable malaria diagnostics. The role of explainable AI techniques in improving clinical adoption and trustworthiness is also underscored. By addressing these challenges, this work advances the field of AI-driven malaria detection and provides actionable insights for researchers and practitioners. The proposed solutions aim to support the development of accessible and accurate diagnostic tools, particularly for resource-constrained populations.

EXplainable Artificial Intelligence (XAI) approaches are widely applied for identifying fairness issues in Artificial Intelligence (AI) systems. However, in the context of facial analysis, existing XAI approaches, such as pixel attribution methods, offer explanations for individual images, posing challenges in assessing the overall behavior of a model, which would require labor-intensive manual inspection of a very large number of instances and leaving to the human the task of drawing a general impression of the model behavior from the individual outputs. Addressing this limitation, we introduce FaceX, the first method that provides a comprehensive understanding of face attribute classifiers through summary model explanations. Specifically, FaceX leverages the presence of distinct regions across all facial images to compute a region-level aggregation of model activations, allowing for the visualization of the model's region attribution across 19 predefined regions of interest in facial images, such as hair, ears, or skin. Beyond spatial explanations, FaceX enhances interpretability by visualizing specific image patches with the highest impact on the model's decisions for each facial region within a test benchmark. Through extensive evaluation in various experimental setups, including scenarios with or without intentional biases and mitigation efforts on four benchmarks, namely CelebA, FairFace, CelebAMask-HQ, and Racial Faces in the Wild, FaceX demonstrates high effectiveness in identifying the models' biases.