Research Papers

Explainable Artificial Intelligence (XAI) aims to provide transparent insights into machine learning models, yet the reliability of many feature attribution methods remains a critical challenge. Prior research (Haufe et al., 2014; Wilming et al., 2022, 2023) has demonstrated that these methods often erroneously assign significant importance to non-informative variables, such as suppressor variables, leading to fundamental misinterpretations. Since statistical suppression is induced by feature dependencies, this study investigates whether data whitening, a common preprocessing technique for decorrelation, can mitigate such errors. Using the established XAI-TRIS benchmark (Clark et al., 2024b), which offers synthetic ground-truth data and quantitative measures of explanation correctness, we empirically evaluate 16 popular feature attribution methods applied in combination with 5 distinct whitening transforms. Additionally, we analyze a minimal linear two-dimensional classification problem (Wilming et al., 2023) to theoretically assess whether whitening can remove the impact of suppressor features from Bayes-optimal models. Our results indicate that, while specific whitening techniques can improve explanation performance, the degree of improvement varies substantially across XAI methods and model architectures. These findings highlight the complex relationship between data non-linearities, preprocessing quality, and attribution fidelity, underscoring the vital role of pre-processing techniques in enhancing model interpretability.

The integration of artificial intelligence (AI) into science education is transforming the design and function of learning materials, offering new affordances for personalization, authenticity, and accessibility. This chapter examines how AI technologies are transforming science learning materials across six interrelated domains: 1) integrating AI into scientific practice, 2) enabling adaptive and personalized instruction, 3) facilitating interactive simulations, 4) generating multimodal content, 5) enhancing accessibility for diverse learners, and 6) promoting co-creation through AI-supported content development. These advancements enable learning materials to more accurately reflect contemporary scientific practice, catering to the diverse needs of students. For instance, AI support can enable students to engage in dynamic simulations, interact with real-time data, and explore science concepts through multimodal representations. Educators are increasingly collaborating with generative AI tools to develop timely and culturally responsive instructional resources. However, these innovations also raise critical ethical and pedagogical concerns, including issues of algorithmic bias, data privacy, transparency, and the need for human oversight. To ensure equitable and meaningful science learning, we emphasize the importance of designing AI-supported materials with careful attention to scientific integrity, inclusivity, and student agency. This chapter advocates for a responsible, ethical, and reflective approach to leveraging AI in science education, framing it as a catalyst for innovation while upholding core educational values.

Machine learning (ML) models are increasingly deployed in cybersecurity applications such as phishing detection and network intrusion prevention. However, these models remain vulnerable to adversarial perturbations small, deliberate input modifications that can degrade detection accuracy and compromise interpretability. This paper presents an empirical study of adversarial robustness and explainability drift across two cybersecurity domains phishing URL classification and network intrusion detection. We evaluate the impact of L (infinity) bounded Fast Gradient Sign Method (FGSM) and Projected Gradient Descent (PGD) perturbations on model accuracy and introduce a quantitative metric, the Robustness Index (RI), defined as the area under the accuracy perturbation curve. Gradient based feature sensitivity and SHAP based attribution drift analyses reveal which input features are most susceptible to adversarial manipulation. Experiments on the Phishing Websites and UNSW NB15 datasets show consistent robustness trends, with adversarial training improving RI by up to 9 percent while maintaining clean-data accuracy. These findings highlight the coupling between robustness and interpretability degradation and underscore the importance of quantitative evaluation in the design of trustworthy, AI-driven cybersecurity systems.

As AI systems increasingly become embedded in interactive and im-mersive artistic environments, artists and technologists are discovering new opportunities to engage with their interpretive and autonomous capacities as creative collaborators in live performance. The focus of this work-in-progress is on outlining conceptual and technical foundations under which performance-makers and interactive architecture can collaborate within rehearsal settings. It introduces a rehearsal-oriented prototype system for shaping and testing AI-mediated environments within creative practice. This approach treats interactive architecture as a performative agent that senses spatial behaviour and speech, interprets these signals through a large language model, and generates real-time environmental adaptations. Designed for deployment in physical performance spaces, the system employs virtual blueprints to support iterative experimentation and creative dialogue between artists and AI agents, using reasoning traces to inform architectural interaction design grounded in dramaturgical principles.

Incentive-Aware AI Safety via Strategic Resource Allocation: A Stackelberg Security Games Perspective

Cheol Woo Kim, Davin Choo, Tzeh Yuan Neoh +1

As AI systems grow more capable and autonomous, ensuring their safety and reliability requires not only model-level alignment but also strategic oversight of the humans and institutions involved in their development and deployment. Existing safety frameworks largely treat alignment as a static optimization problem (e.g., tuning models to desired behavior) while overlooking the dynamic, adversarial incentives that shape how data are collected, how models are evaluated, and how they are ultimately deployed. We propose a new perspective on AI safety grounded in Stackelberg Security Games (SSGs): a class of game-theoretic models designed for adversarial resource allocation under uncertainty. By viewing AI oversight as a strategic interaction between defenders (auditors, evaluators, and deployers) and attackers (malicious actors, misaligned contributors, or worst-case failure modes), SSGs provide a unifying framework for reasoning about incentive design, limited oversight capacity, and adversarial uncertainty across the AI lifecycle. We illustrate how this framework can inform (1) training-time auditing against data/feedback poisoning, (2) pre-deployment evaluation under constrained reviewer resources, and (3) robust multi-model deployment in adversarial environments. This synthesis bridges algorithmic alignment and institutional oversight design, highlighting how game-theoretic deterrence can make AI oversight proactive, risk-aware, and resilient to manipulation.

This paper addresses robotic system engineering for safety- and mission-critical applications by bridging the gap between high-level objectives and formal, executable specifications. The proposed method, Robotic System Task to Model Transformation Methodology (RSTM2) is an ontology-driven, hierarchical approach using stochastic timed Petri nets with resources, enabling Monte Carlo simulations at mission, system, and subsystem levels. A hypothetical case study demonstrates how the RSTM2 method supports architectural trades, resource allocation, and performance analysis under uncertainty. Ontological concepts further enable explainable AI-based assistants, facilitating fully autonomous specification synthesis. The methodology offers particular benefits to complex multi-robot systems, such as the NASA CADRE mission, representing decentralized, resource-aware, and adaptive autonomous systems of the future.

Machine Unlearning aims to remove the influence of specific data or concepts from trained models while preserving overall performance, a capability increasingly required by data protection regulations and responsible AI practices. Despite recent progress, unlearning in text-to-image diffusion models remains challenging due to high computational costs and the difficulty of balancing effective forgetting with retention of unrelated concepts. We introduce Self-distillation for PARameter Efficient Removal (SPARE), a two-stage unlearning method for image generation that combines parameter localization with self-distillation. SPARE first identifies parameters most responsible for generation of the unwanted concepts using gradient-based saliency and constrains updates through sparse low rank adapters, ensuring lightweight, localized modifications. In a second stage, SPARE applies a self-distillation objective that overwrites the unwanted concept with a user-defined surrogate while preserving behavior for other concepts. In addition we proposed a timestep sampling scheme for diffusion models to target only the crucial timesteps for a given concept leading to efficient unlearning. SPARE surpasses the current state-of-the-art on the UnlearnCanvas benchmark, and ablation studies on several datasets indicate fine-grained control over the forgetting-retention trade-off. Our results demonstrate that SPARE achieves strong concept erasure and high retainability across various domains, making it a suitable solution for selective unlearning in diffusion-based image generation models.

With the rapid growth of Large Language Models (LLMs), criticism of their societal impact has also grown. Work in Responsible AI (RAI) has focused on the development of AI systems aimed at reducing harm. Responding to RAI's criticisms and the need to bring the wisdom traditions into HCI, we apply Conwill et al.'s Virtue-Guided Technology Design method to LLMs. We cataloged new ethical design patterns for LLMs and evaluated them through interviews with technologists. Participants valued that the patterns provided more accuracy and robustness, better safety, new research opportunities, increased access and control, and reduced waste. Their concerns were that the patterns could be vulnerable to jailbreaking, were generalizing models too widely, and had potential implementation issues. Overall, participants reacted positively while also acknowledging the tradeoffs involved in ethical LLM design.

Accurate dental diagnosis is essential for oral healthcare, yet many individuals lack access to timely professional evaluation. Existing AI-based methods primarily treat diagnosis as a visual pattern recognition task and do not reflect the structured clinical reasoning used by dental professionals. These approaches also require large amounts of expert-annotated data and often struggle to generalize across diverse real-world imaging conditions. To address these limitations, we present OMNI-Dent, a data-efficient and explainable diagnostic framework that incorporates clinical reasoning principles into a Vision-Language Model (VLM)-based pipeline. The framework operates on multi-view smartphone photographs,embeds diagnostic heuristics from dental experts, and guides a general-purpose VLM to perform tooth-level evaluation without dental-specific fine-tuning of the VLM. By utilizing the VLM's existing visual-linguistic capabilities, OMNI-Dent aims to support diagnostic assessment in settings where curated clinical imaging is unavailable. Designed as an early-stage assistive tool, OMNI-Dent helps users identify potential abnormalities and determine when professional evaluation may be needed, offering a practical option for individuals with limited access to in-person care.

What LLMs Think When You Don't Tell Them What to Think About?

Yongchan Kwon, James Zou

Characterizing the behavior of large language models (LLMs) across diverse settings is critical for reliable monitoring and AI safety. However, most existing analyses rely on topic- or task-specific prompts, which can substantially limit what can be observed. In this work, we study what LLMs generate from minimal, topic-neutral inputs and probe their near-unconstrained generative behavior. Despite the absence of explicit topics, model outputs cover a broad semantic space, and surprisingly, each model family exhibits strong and systematic topical preferences. GPT-OSS predominantly generates programming (27.1%) and mathematical content (24.6%), whereas Llama most frequently generates literary content (9.1%). DeepSeek often generates religious content, while Qwen frequently generates multiple-choice questions. Beyond topical preferences, we also observe differences in content specialization and depth: GPT-OSS often generates more technically advanced content (e.g., dynamic programming) compared with other models (e.g., basic Python). Furthermore, we find that the near-unconstrained generation often degenerates into repetitive phrases, revealing interesting behaviors unique to each model family. For instance, degenerate outputs from Llama include multiple URLs pointing to personal Facebook and Instagram accounts. We release the complete dataset of 256,000 samples from 16 LLMs, along with a reproducible codebase.

Fairness monitoring is critical for detecting algorithmic bias, as mandated by the EU AI Act. Since such monitoring requires sensitive user data (e.g., ethnicity), the AI Act permits its processing only with strict privacy measures, such as multi-party computation (MPC), in compliance with the GDPR. However, the effectiveness of such secure monitoring protocols ultimately depends on people's willingness to share their data. Little is known about how different MPC protocol designs shape user acceptance. To address this, we conducted an online survey with 833 participants in Europe, examining user acceptance of various MPC protocol designs for fairness monitoring. Findings suggest that users prioritized risk-related attributes (e.g., privacy protection mechanism) in direct evaluation but benefit-related attributes (e.g., fairness objective) in simulated choices, with acceptance shaped by their fairness and privacy orientations. We derive implications for deploying and communicating privacy-preserving protocols in ways that foster informed consent and align with user expectations.

Large language model (LLM)-based agents are increasingly used to perform complex, multi-step workflows in regulated settings such as compliance and due diligence. However, many agentic architectures rely primarily on prompt engineering of a single agent, making it difficult to observe or compare how models handle uncertainty and coordination across interconnected decision stages and with human oversight. We introduce a multi-agent system formalized as a finite-horizon Markov Decision Process (MDP) with a directed acyclic structure. Each agent corresponds to a specific role or decision stage (e.g., content, business, or legal review in a compliance workflow), with predefined transitions representing task escalation or completion. Epistemic uncertainty is quantified at the agent level using Monte Carlo estimation, while system-level uncertainty is captured by the MDP's termination in either an automated labeled state or a human-review state. We illustrate the approach through a case study in AI safety evaluation for self-harm detection, implemented as a multi-agent compliance system. Results demonstrate improvements over a single-agent baseline, including up to a 19\% increase in accuracy, up to an 85x reduction in required human review, and, in some configurations, reduced processing time.