Research Papers

Under the slogan of trustworthy AI, much of contemporary AI research is focused on designing AI systems and usage practices that inspire human trust and, thus, enhance adoption of AI systems. However, a person affected by an AI system may not be convinced by AI system design alone -- neither should they, if the AI system is embedded in a social context that gives good reason to believe that it is used in tension with a person's interest. In such cases, distrust in the system may be justified and necessary to build meaningful trust in the first place. We propose the term "healthy distrust" to describe such a justified, careful stance towards certain AI usage practices. We investigate prior notions of trust and distrust in computer science, sociology, history, psychology, and philosophy, outline a remaining gap that healthy distrust might fill and conceptualize healthy distrust as a crucial part for AI usage that respects human autonomy.

How May U.S. Courts Scrutinize Their Recidivism Risk Assessment Tools? Contextualizing AI Fairness Criteria on a Judicial Scrutiny-based Framework

Tin Nguyen, Jiannan Xu, Phuong-Anh Nguyen-Le +4

The AI/HCI and legal communities have developed largely independent conceptualizations of fairness. This conceptual difference hinders the potential incorporation of technical fairness criteria (e.g., procedural, group, and individual fairness) into sustainable policies and designs, particularly for high-stakes applications like recidivism risk assessment. To foster common ground, we conduct legal research to identify if and how technical AI conceptualizations of fairness surface in primary legal sources. We find that while major technical fairness criteria can be linked to constitutional mandates such as ``Due Process'' and ``Equal Protection'' thanks to judicial interpretation, several challenges arise when operationalizing them into concrete statutes/regulations. These policies often adopt procedural and group fairness but ignore the major technical criterion of individual fairness. Regarding procedural fairness, judicial ``scrutiny'' categories are relevant but may not fully capture how courts scrutinize the use of demographic features in potentially discriminatory government tools like RRA. Furthermore, some policies contradict each other on whether to apply procedural fairness to certain demographic features. Thus, we propose a new framework, integrating U.S. demographics-related legal scrutiny concepts and technical fairness criteria, and contextualize it in three other major AI-adopting jurisdictions (EU, China, and India).

Automatically grading the diverse range of question types in high school physics problem is a challenge that requires automated grading techniques from different fields. We report the findings of a Systematic Literature Review of potential physics grading techniques. We propose a multi-modal AI grading framework to address these challenges and examine our framework in light of Australia's AI Ethical Principles.

Advancing AI Research Assistants with Expert-Involved Learning

Tianyu Liu, Simeng Han, Hanchen Wang +27

Large language models (LLMs) and large multimodal models (LMMs) promise to accelerate biomedical discovery, yet their reliability remains unclear. We introduce ARIEL (AI Research Assistant for Expert-in-the-Loop Learning), an open-source evaluation and optimization framework that pairs a curated multimodal biomedical corpus with expert-vetted tasks to probe two capabilities: full-length article summarization and fine-grained figure interpretation. Using uniform protocols and blinded PhD-level evaluation, we find that state-of-the-art models generate fluent but incomplete summaries, whereas LMMs struggle with detailed visual reasoning. We later observe that prompt engineering and lightweight fine-tuning substantially improve textual coverage, and a compute-scaled inference strategy enhances visual question answering. We build an ARIEL agent that integrates textual and visual cues, and we show it can propose testable mechanistic hypotheses. ARIEL delineates current strengths and limitations of foundation models, and provides a reproducible platform for advancing trustworthy AI in biomedicine.

MERA: Multimodal and Multiscale Self-Explanatory Model with Considerably Reduced Annotation for Lung Nodule Diagnosis

Jiahao Lu, Chong Yin, Silvia Ingala +3

Lung cancer, a leading cause of cancer-related deaths globally, emphasises the importance of early detection for better patient outcomes. Pulmonary nodules, often early indicators of lung cancer, necessitate accurate, timely diagnosis. Despite Explainable Artificial Intelligence (XAI) advances, many existing systems struggle providing clear, comprehensive explanations, especially with limited labelled data. This study introduces MERA, a Multimodal and Multiscale self-Explanatory model designed for lung nodule diagnosis with considerably Reduced Annotation requirements. MERA integrates unsupervised and weakly supervised learning strategies (self-supervised learning techniques and Vision Transformer architecture for unsupervised feature extraction) and a hierarchical prediction mechanism leveraging sparse annotations via semi-supervised active learning in the learned latent space. MERA explains its decisions on multiple levels: model-level global explanations via semantic latent space clustering, instance-level case-based explanations showing similar instances, local visual explanations via attention maps, and concept explanations using critical nodule attributes. Evaluations on the public LIDC dataset show MERA's superior diagnostic accuracy and self-explainability. With only 1% annotated samples, MERA achieves diagnostic accuracy comparable to or exceeding state-of-the-art methods requiring full annotation. The model's inherent design delivers comprehensive, robust, multilevel explanations aligned closely with clinical practice, enhancing trustworthiness and transparency. Demonstrated viability of unsupervised and weakly supervised learning lowers the barrier to deploying diagnostic AI in broader medical domains. Our complete code is open-source available: https://github.com/diku-dk/credanno.

Explainability and its emerging counterpart contestability have become important normative and design principles for trustworthy AI as they enable users and subjects to understand and challenge AI decisions. However, realizing these principles is difficult, as they assume different meanings in technical, legal, and organizational dimensions of AI regulation. To resolve this conceptual polysemy, in this paper, we present the findings of an interview study with 14 experts to examine the intersection and implementation of explainability and contestability, and their understanding in different research communities. We outline differentiations between descriptive and normative explainability, judicial and non-judicial channels of contestation, and individual and collective contestation action. We further describe the main points of friction in the realization of both principles, including the alignment between top-down and bottom-up regulation, the assignment of responsibility, and the need for interdisciplinary collaboration. Lastly, we formulate three recommendations for AI policy to implement both principles through a Regulation by Design perspective. We believe our contributions can inform policy-making and regulation of these core principles and enable more effective and equitable design, development, and deployment of trustworthy public AI systems.

Rule-based systems are a very popular form of explainable AI, particularly in the fuzzy community, where fuzzy rules are widely used for control and classification problems. However, fuzzy rule-based classifiers struggle to reach bigger traction outside of fuzzy venues, because users sometimes do not know about fuzzy and because fuzzy partitions are not so easy to interpret in some situations. In this work, we propose a methodology to reduce fuzzy rule-based classifiers to crisp rule-based classifiers. We study different possible crisp descriptions and implement an algorithm to obtain them. Also, we analyze the complexity of the resulting crisp classifiers. We believe that our results can help both fuzzy and non-fuzzy practitioners understand better the way in which fuzzy rule bases partition the feature space and how easily one system can be translated to another and vice versa. Our complexity metric can also help to choose between different fuzzy classifiers based on what the equivalent crisp partitions look like.

Discrimination-aware classification aims to make accurate predictions while satisfying fairness constraints. Traditional decision tree learners typically optimize for information gain in the target attribute alone, which can result in models that unfairly discriminate against protected social groups (e.g., gender, ethnicity). Motivated by these shortcomings, we propose GroupCART, a tree-based ensemble optimizer that avoids bias during model construction by optimizing not only for decreased entropy in the target attribute but also for increased entropy in protected attributes. Our experiments show that GroupCART achieves fairer models without data transformation and with minimal performance degradation. Furthermore, the method supports customizable weighting, offering a smooth and flexible trade-off between predictive performance and fairness based on user requirements. These results demonstrate that algorithmic bias in decision tree models can be mitigated through multi-task, fairness-aware learning. All code and datasets used in this study are available at: https://github.com/anonymous12138/groupCART.

Uncovering the Structure of Explanation Quality with Spectral Analysis

Johannes Maeß, Grégoire Montavon, Shinichi Nakajima +2

As machine learning models are increasingly considered for high-stakes domains, effective explanation methods are crucial to ensure that their prediction strategies are transparent to the user. Over the years, numerous metrics have been proposed to assess quality of explanations. However, their practical applicability remains unclear, in particular due to a limited understanding of which specific aspects each metric rewards. In this paper we propose a new framework based on spectral analysis of explanation outcomes to systematically capture the multifaceted properties of different explanation techniques. Our analysis uncovers two distinct factors of explanation quality-stability and target sensitivity-that can be directly observed through spectral decomposition. Experiments on both MNIST and ImageNet show that popular evaluation techniques (e.g., pixel-flipping, entropy) partially capture the trade-offs between these factors. Overall, our framework provides a foundational basis for understanding explanation quality, guiding the development of more reliable techniques for evaluating explanations.

On Background Bias of Post-Hoc Concept Embeddings in Computer Vision DNNs

Gesina Schwalbe, Georgii Mikriukov, Edgar Heinert +5

The thriving research field of concept-based explainable artificial intelligence (C-XAI) investigates how human-interpretable semantic concepts embed in the latent spaces of deep neural networks (DNNs). Post-hoc approaches therein use a set of examples to specify a concept, and determine its embeddings in DNN latent space using data driven techniques. This proved useful to uncover biases between different target (foreground or concept) classes. However, given that the background is mostly uncontrolled during training, an important question has been left unattended so far: Are/to what extent are state-of-the-art, data-driven post-hoc C-XAI approaches themselves prone to biases with respect to their backgrounds? E.g., wild animals mostly occur against vegetation backgrounds, and they seldom appear on roads. Even simple and robust C-XAI methods might abuse this shortcut for enhanced performance. A dangerous performance degradation of the concept-corner cases of animals on the road could thus remain undiscovered. This work validates and thoroughly confirms that established Net2Vec-based concept segmentation techniques frequently capture background biases, including alarming ones, such as underperformance on road scenes. For the analysis, we compare 3 established techniques from the domain of background randomization on >50 concepts from 2 datasets, and 7 diverse DNN architectures. Our results indicate that even low-cost setups can provide both valuable insight and improved background robustness.

Multimodal prediction models based on imaging and clinical data are increasingly used for clinical decision support, yet their interpretability remains limited. We present multimodal Deep Transformation Models (DTMs) combining statistical approaches and neural networks to achieve strong predictive performance while preserving interpretability for tabular data. A key contribution of this work is the adaption of the xAI methods Grad-CAM and Occlusion to DTMs relying on 3D CNNs, enabling interpretation of the image branch through the generation of explanation maps. We developed DTMs to predict functional independence three months after stroke using diffusion-weighted imaging and clinical data from 407 patients. In a ten-fold cross-validation, the models achieved state-of-the-art predictive performance (AUC 0.81 [0.75, 0.87]) while maintaining interpretability for tabular features, with functional independence before stroke and stroke severity on admission emerging as the strongest predictors. Explanation maps from both xAI methods highlighted consistent regions, including frontal lobe areas which are known to be associated with age, a strong predictor of functional outcome. Notably, these regions disappeared once age was included as an explicit tabular predictor. Similarity analyses of explanation maps revealed distinct spatial patterns, providing meaningful insights into stroke pathophysiology, systematic error analysis and hypothesis generation.

The study of neural networks from the perspective of Fourier features has garnered significant attention. While existing analytical research suggests that neural networks tend to learn low-frequency features, a clear attribution method for identifying the specific learned Fourier features has remained elusive. To bridge this gap, we propose a novel Fourier feature attribution method grounded in signal decomposition theory. Additionally, we analyze the differences between game-theoretic attribution metrics for Fourier and spatial domain features, demonstrating that game-theoretic evaluation metrics are better suited for Fourier-based feature attribution. Our experiments show that Fourier feature attribution exhibits superior feature selection capabilities compared to spatial domain attribution methods. For instance, in the case of Vision Transformers (ViTs) on the ImageNet dataset, only $8\%$ of the Fourier features are required to maintain the original predictions for $80\%$ of the samples. Furthermore, we compare the specificity of features identified by our method against traditional spatial domain attribution methods. Results reveal that Fourier features exhibit greater intra-class concentration and inter-class distinctiveness, indicating their potential for more efficient classification and explainable AI algorithms.