{"id":5685,"date":"2026-02-14T06:24:43","date_gmt":"2026-02-14T06:24:43","guid":{"rendered":"https:\/\/scipapermill.com\/index.php\/2026\/02\/14\/explainable-ai-unveiling-the-why-in-ais-decisions-from-edge-to-agentic-systems\/"},"modified":"2026-02-14T06:24:43","modified_gmt":"2026-02-14T06:24:43","slug":"explainable-ai-unveiling-the-why-in-ais-decisions-from-edge-to-agentic-systems","status":"publish","type":"post","link":"https:\/\/scipapermill.com\/index.php\/2026\/02\/14\/explainable-ai-unveiling-the-why-in-ais-decisions-from-edge-to-agentic-systems\/","title":{"rendered":"Explainable AI: Unveiling the “Why” in AI’s Decisions, from Edge to Agentic Systems"},"content":{"rendered":"

Latest 16 papers on explainable ai: Feb. 14, 2026<\/h3>\n

The quest for transparent and trustworthy AI has never been more pressing. As AI models permeate every facet of our lives, from critical medical diagnostics to autonomous vehicles and even sophisticated software agents, the demand to understand how<\/em> these systems arrive at their decisions has skyrocketed. This digest dives into a fascinating collection of recent breakthroughs in Explainable AI (XAI), showcasing innovations that push the boundaries of interpretability across diverse applications.<\/p>\n

The Big Idea(s) & Core Innovations<\/h3>\n

The overarching theme in recent XAI research is a concerted effort to make AI less of a black box, adapting explanations to varying contexts and user needs. A crucial insight, highlighted by Benedict Clark et al.<\/strong> from University of Cambridge<\/strong> in their paper, \u201cFeature salience \u2013 not task-informativeness \u2013 drives machine learning model explanations\u201d<\/a>, challenges a fundamental assumption: that feature importance is driven by task-relevance. Instead, they demonstrate that feature salience<\/em> (like image structures) can be the primary driver, calling for a critical re-evaluation of existing XAI methods to avoid spurious correlations.<\/p>\n

Building on the need for more relevant explanations, Muhammad Rashid et al.<\/strong> from University of Torino<\/strong> in \u201cShapBPT: Image Feature Attributions Using Data-Aware Binary Partition Trees\u201d<\/a> introduce ShapBPT. This novel method uses hierarchical Shapley values tailored for image data, aligning attributions with intrinsic image morphology for more semantically meaningful visual explanations. Similarly, Vasileios Arampatzakis et al.<\/strong> from Democritus University of Thrace<\/strong> introduce SVDA in \u201cInterpretable Vision Transformers in Image Classification via SVDA\u201d<\/a>, a geometrically grounded attention mechanism for Vision Transformers. SVDA enhances interpretability by introducing spectral and directional constraints, providing structured attention patterns without sacrificing accuracy.<\/p>\n

Beyond visual explanations, Lars H. B. Olsen and Danniel Christensen<\/strong> from the University of Bergen, Norway<\/strong>, in \u201cComputing Conditional Shapley Values Using Tabular Foundation Models\u201d<\/a>, demonstrate how tabular foundation models like TabPFN can efficiently and accurately compute conditional Shapley values, particularly for smooth predictive models. This opens new avenues for interpreting complex models, especially with diverse datasets. In a more practical agricultural context, Alam, B. M. S. et al.<\/strong> propose in \u201cToward Reliable Tea Leaf Disease Diagnosis Using Deep Learning Model: Enhancing Robustness With Explainable AI and Adversarial Training\u201d<\/a> that integrating XAI techniques like Grad-CAM with adversarial training not only improves interpretability in tea leaf disease diagnosis but also significantly enhances model robustness against noise.<\/p>\n

Crucially, the scope of XAI is expanding beyond static models. S. Chaduvula et al.<\/strong> from the Vector Institute<\/strong> address a significant gap in \u201cFrom Features to Actions: Explainability in Traditional and Agentic AI Systems\u201d<\/a>, arguing that traditional XAI methods are insufficient for understanding complex, multi-step agentic AI systems<\/em> (like LLM-based agents). They propose a shift towards trajectory-level analysis<\/em>, emphasizing the need to explain sequences of decisions rather than just static feature attributions.<\/p>\n

Under the Hood: Models, Datasets, & Benchmarks<\/h3>\n

Recent XAI research leverages and develops a variety of models, datasets, and benchmarks to validate and demonstrate innovations:<\/p>\n