Research Papers

The success of recent Artificial Intelligence (AI) models has been accompanied by the opacity of their internal mechanisms, due notably to the use of deep neural networks. In order to understand these internal mechanisms and explain the output of these AI models, a set of methods have been proposed, grouped under the domain of eXplainable AI (XAI). This paper focuses on a sub-domain of XAI, called eXplainable Reinforcement Learning (XRL), which aims to explain the actions of an agent that has learned by reinforcement learning. We propose an intuitive taxonomy based on two questions "What" and "How". The first question focuses on the target that the method explains, while the second relates to the way the explanation is provided. We use this taxonomy to provide a state-of-the-art review of over 250 papers. In addition, we present a set of domains close to XRL, which we believe should get attention from the community. Finally, we identify some needs for the field of XRL.

A Different Approach to AI Safety: Proceedings from the Columbia Convening on Openness in Artificial Intelligence and AI Safety

Camille François, Ludovic Péran, Ayah Bdeir +17

The rapid rise of open-weight and open-source foundation models is intensifying the obligation and reshaping the opportunity to make AI systems safe. This paper reports outcomes from the Columbia Convening on AI Openness and Safety (San Francisco, 19 Nov 2024) and its six-week preparatory programme involving more than forty-five researchers, engineers, and policy leaders from academia, industry, civil society, and government. Using a participatory, solutions-oriented process, the working groups produced (i) a research agenda at the intersection of safety and open source AI; (ii) a mapping of existing and needed technical interventions and open source tools to safely and responsibly deploy open foundation models across the AI development workflow; and (iii) a mapping of the content safety filter ecosystem with a proposed roadmap for future research and development. We find that openness -- understood as transparent weights, interoperable tooling, and public governance -- can enhance safety by enabling independent scrutiny, decentralized mitigation, and culturally plural oversight. However, significant gaps persist: scarce multimodal and multilingual benchmarks, limited defenses against prompt-injection and compositional attacks in agentic systems, and insufficient participatory mechanisms for communities most affected by AI harms. The paper concludes with a roadmap of five priority research directions, emphasizing participatory inputs, future-proof content filters, ecosystem-wide safety infrastructure, rigorous agentic safeguards, and expanded harm taxonomies. These recommendations informed the February 2025 French AI Action Summit and lay groundwork for an open, plural, and accountable AI safety discipline.

Large language models (LLMs) deployed as agents introduce significant safety risks in clinical settings due to their potential for error and single points of failure. We introduce Tiered Agentic Oversight (TAO), a hierarchical multi-agent system that enhances AI safety through layered, automated supervision. Inspired by clinical hierarchies (e.g., nurse-physician-specialist) in hospital, TAO routes tasks to specialized agents based on complexity, creating a robust safety framework through automated inter- and intra-tier communication and role-playing. Crucially, this hierarchical structure functions as an effective error-correction mechanism, absorbing up to 24% of individual agent errors before they can compound. Our experiments reveal TAO outperforms single-agent and other multi-agent systems on 4 out of 5 healthcare safety benchmarks, with up to an 8.2% improvement. Ablation studies confirm key design principles of the system: (i) its adaptive architecture is over 3% safer than static, single-tier configurations, and (ii) its lower tiers are indispensable, as their removal causes the most significant degradation in overall safety. Finally, we validated the system's synergy with human doctors in a user study where a physician, acting as the highest tier agent, provided corrective feedback that improved medical triage accuracy from 40% to 60%. Project Page: https://tiered-agentic-oversight.github.io/

Financial literacy is increasingly dependent on the ability to interpret complex financial data and utilize advanced forecasting tools. In this context, this study proposes a novel approach that combines transformer-based time series models with explainable artificial intelligence (XAI) to enhance the interpretability and accuracy of stock price predictions. The analysis focuses on the daily stock prices of the five highest-volume banks listed in the BIST100 index, along with XBANK and XU100 indices, covering the period from January 2015 to March 2025. Models including DLinear, LTSNet, Vanilla Transformer, and Time Series Transformer are employed, with input features enriched by technical indicators. SHAP and LIME techniques are used to provide transparency into the influence of individual features on model outputs. The results demonstrate the strong predictive capabilities of transformer models and highlight the potential of interpretable machine learning to empower individuals in making informed investment decisions and actively engaging in financial markets.

As Artificial Intelligence (AI) systems, particularly those based on machine learning (ML), become integral to high-stakes applications, their probabilistic and opaque nature poses significant challenges to traditional verification and validation methods. These challenges are exacerbated in regulated sectors requiring tamper-proof, auditable evidence, as highlighted by apposite legal frameworks, e.g., the EU AI Act. Conversely, Zero-Knowledge Proofs (ZKPs) offer a cryptographic solution that enables provers to demonstrate, through verified computations, adherence to set requirements without revealing sensitive model details or data. Through a systematic survey of ZKP protocols, we identify five key properties (non-interactivity, transparent setup, standard representations, succinctness, and post-quantum security) critical for their application in AI validation and verification pipelines. Subsequently, we perform a follow-up systematic survey analyzing ZKP-enhanced ML applications across an adaptation of the Team Data Science Process (TDSP) model (Data & Preprocessing, Training & Offline Metrics, Inference, and Online Metrics), detailing verification objectives, ML models, and adopted protocols. Our findings indicate that current research on ZKP-Enhanced ML primarily focuses on inference verification, while the data preprocessing and training stages remain underexplored. Most notably, our analysis identifies a significant convergence within the research domain toward the development of a unified Zero-Knowledge Machine Learning Operations (ZKMLOps) framework. This emerging framework leverages ZKPs to provide robust cryptographic guarantees of correctness, integrity, and privacy, thereby promoting enhanced accountability, transparency, and compliance with Trustworthy AI principles.

People increasingly rely on AI-advice when making decisions. At times, such advice can promote selfish behavior. When individuals abide by selfishness-promoting AI advice, how are they perceived and punished? To study this question, we build on theories from social psychology and combine machine-behavior and behavioral economic approaches. In a pre-registered, financially-incentivized experiment, evaluators could punish real decision-makers who (i) received AI, human, or no advice. The advice (ii) encouraged selfish or prosocial behavior, and decision-makers (iii) behaved selfishly or, in a control condition, behaved prosocially. Evaluators further assigned responsibility to decision-makers and their advisors. Results revealed that (i) prosocial behavior was punished very little, whereas selfish behavior was punished much more. Focusing on selfish behavior, (ii) compared to receiving no advice, selfish behavior was penalized more harshly after prosocial advice and more leniently after selfish advice. Lastly, (iii) whereas selfish decision-makers were seen as more responsible when they followed AI compared to human advice, punishment between the two advice sources did not vary. Overall, behavior and advice content shape punishment, whereas the advice source does not.

Rice is an essential staple food worldwide that is important in promoting international trade, economic growth, and nutrition. Asian countries such as China, India, Pakistan, Thailand, Vietnam, and Indonesia are notable for their significant contribution to the cultivation and utilization of rice. These nations are also known for cultivating different rice grains, including short and long grains. These sizes are further classified as basmati, jasmine, kainat saila, ipsala, arborio, etc., catering to diverse culinary preferences and cultural traditions. For both local and international trade, inspecting and maintaining the quality of rice grains to satisfy customers and preserve a country's reputation is necessary. Manual quality check and classification is quite a laborious and time-consuming process. It is also highly prone to mistakes. Therefore, an automatic solution must be proposed for the effective and efficient classification of different varieties of rice grains. This research paper presents an automatic framework based on a convolutional neural network (CNN) for classifying different varieties of rice grains. We evaluated the proposed model based on performance metrics such as accuracy, recall, precision, and F1-Score. The CNN model underwent rigorous training and validation, achieving a remarkable accuracy rate and a perfect area under each class's Receiver Operating Characteristic (ROC) curve. The confusion matrix analysis confirmed the model's effectiveness in distinguishing between the different rice varieties, indicating minimal misclassifications. Additionally, the integration of explainability techniques such as LIME (Local Interpretable Model-agnostic Explanations) and SHAP (SHapley Additive exPlanations) provided valuable insights into the model's decision-making process, revealing how specific features of the rice grains influenced classification outcomes.

Real-world machine learning models require rigorous evaluation before deployment, especially in safety-critical domains like autonomous driving and surveillance. The evaluation of machine learning models often focuses on data slices, which are subsets of the data that share a set of characteristics. Data slice finding automatically identifies conditions or data subgroups where models underperform, aiding developers in mitigating performance issues. Despite its popularity and effectiveness, data slicing for vision model validation faces several challenges. First, data slicing often needs additional image metadata or visual concepts, and falls short in certain computer vision tasks, such as object detection. Second, understanding data slices is a labor-intensive and mentally demanding process that heavily relies on the expert's domain knowledge. Third, data slicing lacks a human-in-the-loop solution that allows experts to form hypothesis and test them interactively. To overcome these limitations and better support the machine learning operations lifecycle, we introduce VISLIX, a novel visual analytics framework that employs state-of-the-art foundation models to help domain experts analyze slices in computer vision models. Our approach does not require image metadata or visual concepts, automatically generates natural language insights, and allows users to test data slice hypothesis interactively. We evaluate VISLIX with an expert study and three use cases, that demonstrate the effectiveness of our tool in providing comprehensive insights for validating object detection models.

Explainable Artificial Intelligence (XAI) techniques for interpreting object detection models remain in an early stage, with no established standards for systematic evaluation. This absence of consensus hinders both the comparative analysis of methods and the informed selection of suitable approaches. To address this gap, we introduce the Object Detection Explainable AI Evaluation (ODExAI), a comprehensive framework designed to assess XAI methods in object detection based on three core dimensions: localization accuracy, faithfulness to model behavior, and computational complexity. We benchmark a set of XAI methods across two widely used object detectors (YOLOX and Faster R-CNN) and standard datasets (MS-COCO and PASCAL VOC). Empirical results demonstrate that region-based methods (e.g., D-CLOSE) achieve strong localization (PG = 88.49%) and high model faithfulness (OA = 0.863), though with substantial computational overhead (Time = 71.42s). On the other hand, CAM-based methods (e.g., G-CAME) achieve superior localization (PG = 96.13%) and significantly lower runtime (Time = 0.54s), but at the expense of reduced faithfulness (OA = 0.549). These findings demonstrate critical trade-offs among existing XAI approaches and reinforce the need for task-specific evaluation when deploying them in object detection pipelines. Our implementation and evaluation benchmarks are publicly available at: https://github.com/Analytics-Everywhere-Lab/odexai.

Explainable artificial intelligence (XAI) has become increasingly important in decision-critical domains such as healthcare, finance, and law. Counterfactual (CF) explanations, a key approach in XAI, provide users with actionable insights by suggesting minimal modifications to input features that lead to different model outcomes. Despite significant advancements, existing CF generation methods often struggle to balance proximity, diversity, and robustness, limiting their real-world applicability. A widely adopted framework, Diverse Counterfactual Explanations (DiCE), emphasizes diversity but lacks robustness, making CF explanations sensitive to perturbations and domain constraints. To address these challenges, we introduce DiCE-Extended, an enhanced CF explanation framework that integrates multi-objective optimization techniques to improve robustness while maintaining interpretability. Our approach introduces a novel robustness metric based on the Dice-Sørensen coefficient, enabling stability under small input variations. Additionally, we refine CF generation using weighted loss components (lambda_p, lambda_d, lambda_r) to balance proximity, diversity, and robustness. We empirically validate DiCE-Extended on benchmark datasets (COMPAS, Lending Club, German Credit, Adult Income) across multiple ML backends (Scikit-learn, PyTorch, TensorFlow). Results demonstrate improved CF validity, stability, and alignment with decision boundaries compared to standard DiCE-generated explanations. Our findings highlight the potential of DiCE-Extended in generating more reliable and interpretable CFs for high-stakes applications. Future work could explore adaptive optimization techniques and domain-specific constraints to further enhance CF generation in real-world scenarios

This study presents the design and development of the 21st Century Teacher Educator for Ghana GPT, a customized Generative AI (GenAI) tool created using OpenAI's Retrieval-Augmented Generation (RAG) and Interactive Semi-Automated Prompting Strategy (ISA). Anchored in a Glocalized design approach, this tool supports pre-service teachers (PSTs) in Ghana by embedding localized linguistic, cultural, and curricular content within globally aligned principles of ethical and responsible AI use. The model utilizes structured, preloaded datasets-including Ghana's National Teacher Education Curriculum Framework (NTECF), UNESCO's (2023) AI guidelines, and culturally responsive pedagogies-to offer curriculum-aligned, linguistically adaptive, and pedagogically grounded learning support. The ISA enables users to input their institution, year, and semester, generating tailored academic content such as lecture notes, assessment practice, practicum resources, and action research guidance. The design incorporates the Culture and Context-Aware Framework, GenAI-CRSciA, and frameworks addressing GenAI neocolonialism to ensure equity, curriculum fidelity, and local relevance. Pilot implementation revealed notable strengths in language adaptation and localization, delivering bilingual support in English and Ghanaian languages like Twi, Dagbani, Mampruli, and Dagaare, with contextualized examples for deeper understanding. The GPT also generated practice assessments aligned with course objectives, reinforcing learner engagement. Challenges included occasional hallucinations due to limited corpora in some indigenous languages and access barriers tied to premium subscriptions. This design case contributes to discourse on Glocalized GenAI and calls for collaboration with OpenAI NextGen to expand access and empirically assess usage across diverse African educational contexts.

The rapid growth of artificial intelligence (AI) technologies has raised major privacy and ethical concerns. However, existing AI incident taxonomies and guidelines lack grounding in real-world cases, limiting their effectiveness for prevention and mitigation. We analyzed 202 real-world AI privacy and ethical incidents to develop a taxonomy that classifies them across AI lifecycle stages and captures contributing factors, including causes, responsible entities, sources of disclosure, and impacts. Our findings reveal widespread harms from poor organizational decisions and legal non-compliance, limited corrective interventions, and rare reporting from AI developers and adopting entities. Our taxonomy offers a structured approach for systematic incident reporting and emphasizes the weaknesses of current AI governance frameworks. Our findings provide actionable guidance for policymakers and practitioners to strengthen user protections, develop targeted AI policies, enhance reporting practices, and foster responsible AI governance and innovation, especially in contexts such as social media and child protection.