Research Papers

Explainable AI (XAI) methods are commonly evaluated using functional correctness metrics, sometimes termed faithfulness or fidelity, which estimate how closely an explanation reflects the model's reasoning. Higher correctness is assumed to produce better human understanding, but this link has not been tested with controlled levels. We conducted a user study (N=200) that manipulated explanation correctness at four levels (100%, 85%, 70%, 55%) in a synthetic time series classification task where participants could not rely on domain knowledge or visual intuition. Correctness was defined against a known ground truth, not estimated from a trained model. Participants predicted a simulated AI's decisions from feature-attribution-style explanations (forward simulation), and we used their forward simulation accuracy as a proxy for understanding. Correctness affected understanding, but not at every level: forward simulation accuracy dropped at 70% and 55% relative to fully correct explanations, with no further reduction below 70% and no conclusive difference between 85% and 100%. Lower correctness reduced how many participants predicted accurately rather than lowering accuracy uniformly, and even fully correct explanations did not guarantee understanding: forward simulation accuracy was bimodal there, with only a subset of participants well above chance. In exploratory analyses, some participants predicted decisions accurately but described the wrong pattern when asked what the AI used, and self-reported ratings correlated with forward simulation accuracy only when explanations were fully correct. These findings show that not all differences in functional correctness translate to differences in human understanding, highlighting the need to validate functional metrics against human outcomes.

Deep learning (DL) has been widely studied for assisting applications of modern wireless communications. One of the applications is automatic modulation classification (AMC). However, DL models are found to be vulnerable to adversarial machine learning (AML) threats. One of the most persistent and stealthy threats is the backdoor (Trojan) attack. Nevertheless, most studied threats originate from other AI domains, such as computer vision (CV). Therefore, in this paper, a physical backdoor attack targeting the wireless signal before transmission is studied. The adversary is considered to be using explainable AI (XAI) to guide the placement of the trigger in the most vulnerable parts of the signal. Then, a class prototype combined with principal components is used to generate the trigger. The studied threat was found to be efficient in breaching multiple DL-based AMC models. The attack achieves high success rates for a wide range of SNR values and a small poisoning ratio.

Trustworthy artificial intelligence increasingly relies on probabilistic computation to achieve robustness, interpretability, security and privacy. In practical systems, such workloads interleave deterministic data access with repeated stochastic sampling across models, data paths and system functions, shifting performance bottlenecks from arithmetic units to memory systems that must deliver both data and randomness. Here we present a unified data-access perspective in which deterministic access is treated as a limiting case of stochastic sampling, enabling both modes to be analyzed within a common framework. This view reveals that increasing stochastic demand reduces effective data-access efficiency and can drive systems into entropy-limited operation. Based on this insight, we define memory-level evaluation criteria, including unified operation, distribution programmability, efficiency, robustness to hardware non-idealities and parallel compatibility. Using these criteria, we analyze limitations of conventional architectures and examine emerging probabilistic compute-in-memory approaches that integrate sampling with memory access, outlining pathways toward scalable hardware for trustworthy AI.

The aim of this article is to understand the problem of "black box" algorithms, an issue inherent to the nascent field of Explainable Artificial Intelligence (XAI). While it is relatively easy to understand something someone explained to us, it becomes more complicated when no one can fully grasp the issue. Our purpose is however to highlight: (1) that we should speak of interpretability rather than explainability when we seek to understand models, mainly because we never have complete and unambiguous access to information; (2) that the machines face the problem of the inscrutability of reference, in the same way that the linguist imagined by Willard Van Orman Quine cannot precisely determine what the term "gavagai" refers to in a situation of radical translation; (3) that there is no rule for the application of language, except for "language games", as Ludwig Wittgenstein's linguistics teaches us. The hope of achieving complete explicability and transparency of algorithms is undoubtedly in vain: we can only rely on partial and broad interpretations that will never fully explain the underlying rules.

We present the Consilium Protocol, a Byzantine Fault Tolerance-derived architecture for structured multi-model AI deliberation that treats inter-model disagreement as epistemic signal rather than error. The protocol assigns engineered cognitive personas to language models -- separating what a model is from how it reasons -- and introduces an In-Sample/Out-of-Sample validation framework adapted from quantitative finance to distinguish training-data consensus from empirically grounded conclusions. Across 1,478 deliberation sessions spanning 32 topics in 10 domain categories, we demonstrate that (1) the cognitive persona, not the underlying model, determines epistemic behavior: free edge-inference models costing 0.0002 USD per batch produced comparable analytical output to frontier models costing 10.69 USD; (2) RLHF alignment training creates measurable, domain-specific epistemic blind spots -- contested policy topics exhibit 12.3 percentage points less adversarial challenge than settled science topics, and AI safety topics show asymmetric bias ($Δ$=11.6%) where models challenge claims that AI is dangerous far more vigorously than claims that AI risk is overstated; (3) the protocol exhibits no directional bias of its own (immigration $Δ$=2.3%, renewables $Δ$=1.2%); and (4) out-of-sample evidence retrieval validated 239 claims with 100% evidence retrieval and surfaced 167 blind-spot discoveries invisible to training-data deliberation. Run-to-run reproducibility across randomized model$\times$persona assignments averages $\pm$2.2% standard deviation. Total cost for the complete battery including all overhead: 217 USD. We release the protocol specification under MIT license to enable independent verification.

The first generation of students is learning to program alongside GenAI (Generative Artificial Intelligence) tools, raising questions about how young learners critically engage with them and perceive ethical responsibilities. While prior research has focused on university students or developers, little is known about secondary school novices, who represent the next cohort of software engineers. To address this gap, we conducted an exploratory study with 84 German secondary school students aged 16-19 attending software development workshops. We examined their critical thinking practices in AI-assisted programming, perceptions of AI ethics and responsibility, and gender-related differences in their views. Our results reveal an AI paradox: students demonstrate strong ethical reasoning and awareness about AI, yet many report integrating AI-generated code without a thorough understanding of it. The majority of our cohort attributed significant responsibility for AI practices to politics and corporations, potentially reflecting Germany's cultural context, with its strict regulations and data privacy discourse. Boys reported more frequent and experimental use of AI-assisted programming, whereas girls expressed greater scepticism and emphasised peer collaboration over GenAI assistance. Our findings highlight the importance of culturally responsive software engineering education that strengthens critical AI literacy in AI-assisted programming by linking ethics to concrete code artefacts and preparing young learners for this AI-driven software landscape.

Purpose: Understanding a technical artifact requires grasping both its internal structure (Architecture) and its purpose and significance (Relevance), as formalized by Dual Nature Theory. This controlled experimental study investigates whether how explainers address these perspectives affects explainees' understanding. Methods: In a between-subjects experiment, 104 participants received explanations of the board game Quarto! from trained confederates in one of three conditions: Architecture-focused (A), Relevance-focused (R), or Integrated (AR). Understanding was assessed on comprehension (knowing that) and enabledness (knowing how). Results: The A and R conditions produced equivalent understanding despite different explanation content. The AR condition yielded significantly higher enabledness than the focused conditions combined $\mathrm{F}(1, 102) = 4.83$, $p = .030$, $η^2_p = .045$}, while no differences emerged for comprehension. Conclusion: Integrating Architecture and Relevance specifically enhances explainees' ability to apply their understanding in practice, suggesting that fostering agency with technical artifacts requires bridging both perspectives. This has implications for technology education and explainable AI design.

Research on explainable AI (XAI) has frequently focused on explaining model predictions. More recently, methods have been proposed to explain prediction uncertainty by attributing it to input features (uncertainty attributions). However, the evaluation of these methods remains inconsistent as studies rely on heterogeneous proxy tasks and metrics, hindering comparability. We address this by aligning uncertainty attributions with the well-established Co-12 framework for XAI evaluation. We propose concrete implementations for the correctness, consistency, continuity, and compactness properties. Additionally, we introduce conveyance, a property tailored to uncertainty attributions that evaluates whether controlled increases in epistemic uncertainty reliably propagate to feature-level attributions. We demonstrate our evaluation framework with eight metrics across combinations of uncertainty quantification and feature attribution methods on tabular and image data. Our experiments show that gradient-based methods consistently outperform perturbation-based approaches in consistency and conveyance, while Monte-Carlo dropconnect outperforms Monte-Carlo dropout in most metrics. Although most metrics rank the methods consistently across samples, inter-method agreement remains low. This suggests no single metric sufficiently evaluates uncertainty attribution quality. The proposed evaluation framework contributes to the body of knowledge by establishing a foundation for systematic comparison and development of uncertainty attribution methods.

Dissecting Model Failures in Abdominal Aortic Aneurysm Segmentation through Explainability-Driven Analysis

Abu Noman Md Sakib, Merjulah Roby, Zijie Zhang +3

Computed tomography image segmentation of complex abdominal aortic aneurysms (AAA) often fails because the models assign internal focus to irrelevant structures or do not focus on thin, low-contrast targets. Where the model looks is the primary training signal, and thus we propose an Explainable AI (XAI) guided encoder shaping framework. Our method computes a dense, attribution-based encoder focus map ("XAI field") from the final encoder block and uses it in two complementary ways: (i) we align the predicted probability mass to the XAI field to promote agreement between focus and output; and (ii) we route the field into a lightweight refinement pathway and a confidence prior that modulates logits at inference, suppressing distractors while preserving subtle structures. The objective terms serve only as control signals; the contribution is the integration of attribution guidance into representation and decoding. We evaluate clinically validated challenging cases curated for failure-prone scenarios. Compared to a base SAM setup, our implementation yields substantial improvements. The observed gains suggest that explicitly optimizing encoder focus via XAI guidance is a practical and effective principle for reliable segmentation in complex scenarios.

Responsible AI initiatives place great emphasis on the safety of Large Language Model (LLM)-based systems. In particular, it has become standard practice to subject these models to an alignment procedure aimed at preventing harmful outputs. However, once aligned, a model is not guaranteed to maintain this alignment throughout its lifecycle. Moreover, the likelihood of misalignment increases as malicious actors may deliberately employ jailbreaking techniques to compromise LLM safety. To counter this, much research has focused on improving alignment methods and post-processing filters. In this paper, we introduce a new perspective on advancing LLM alignment: rather than developing stronger alignment techniques, we investigate the model's intrinsic ability to recover its alignment after corruption. We propose a methodology for modeling the safety trajectories of user-assistant interactions and for detecting recovery trends within them. We apply this approach to a jailbreaking scenario, presenting a preliminary recovery analysis based on a dataset of adversarial multi-turn dialogues and examining the influence of the content moderation model chosen for safety evaluation. Project page with an interactive data visualizer is available at https://lab-rococo-sapienza.github.io/LearningfromMistakes.

The advancing fluency of LLMs raises important questions about their ability to emulate complex human traits, including emotional expression and personality, across diverse linguistic and cultural contexts. This study investigates whether LLMs can convincingly mimic emotional nuance in English and personality markers in Arabic, a critical under-resourced language with unique linguistic and cultural characteristics. We conduct two tasks across six models:Jais, Mistral, LLaMA, GPT-4o, Gemini, and DeepSeek. First, we evaluate whether machine classifiers can reliably distinguish between human-authored and AI-generated texts. Second, we assess the extent to which LLM-generated texts exhibit emotional or personality traits comparable to those of humans. Our results demonstrate that AI-generated texts are distinguishable from human-authored ones (F1>0.95), though classification performance deteriorates on paraphrased samples, indicating a reliance on superficial stylistic cues. Emotion and personality classification experiments reveal significant generalization gaps: classifiers trained on human data perform poorly on AI-generated texts and vice versa, suggesting LLMs encode affective signals differently from humans. Importantly, augmenting training with AI-generated data enhances performance in the Arabic personality classification task, highlighting the potential of synthetic data to address challenges in under-resourced languages. Model-specific analyses show that GPT-4o and Gemini exhibit superior affective coherence. Linguistic and psycholinguistic analyses reveal measurable divergences in tone, authenticity, and textual complexity between human and AI texts. These findings have implications for affective computing, authorship attribution, and responsible AI deployment, particularly within underresourced language contexts where generative AI detection and alignment pose unique challenges.

Reliable photovoltaic defect identification is essential for maintaining energy yield, ensuring warranty compliance, and enabling scalable inspection of rapidly expanding solar fleets. Although recent advances in computer vision have improved automated defect detection, most existing systems operate as opaque classifiers that provide limited diagnostic insight for high-stakes energy infrastructure. Here we introduce REVL-PV, a vision-language framework that embeds domain-specific diagnostic reasoning into multimodal learning across electroluminescence, thermal, and visible-light imagery. By requiring the model to link visual evidence to plausible defect mechanisms before classification, the framework produces structured diagnostic reports aligned with professional photovoltaic inspection practice. Evaluated on 1,927 real-world modules spanning eight defect categories, REVL-PV achieves 93\% classification accuracy while producing interpretable diagnostic rationales and maintaining strong robustness under realistic image corruptions. A blind concordance study with a certified solar inspection expert shows strong semantic alignment between model explanations and expert assessments across defect identification, root-cause attribution, and visual descriptions. These results demonstrate that reasoning-aware multimodal learning establishes a general paradigm for trustworthy AI-assisted inspection of photovoltaic energy infrastructure.