Explainable and Trustworthy Artificial Intelligence is gaining significant attention from the scientific community. The session aims to explore and disseminate novel methods in explainability that improve transparency and user understanding of complex AI models. Investigating methods for ensuring AI systems adhere to principles of fairness, accountability, and robustness fosters trust in AI systems and enables real-world applications. The EXTAIS 2025 Special Session at the 17th International Conference on Computational Collective Intelligence (ICCCI 2025) is devoted to methods and techniques that allow humans to understand, trust, and interact effectively with AI systems. Clarifying how AI models make decisions, interpret predictions, and function in complex environments enhances AI transparency, accountability, and reliability.
The scope of the session includes, but is not limited to, the following topics:
- Theoretical frameworks and definitions of explainability in AI
- Cognitive science approaches to AI interpretability
- Fairness, accountability, transparency, and ethics in AI
- Legal, ethical, and societal implications of AI transparency
- Techniques for explainability in neural networks, decision trees, ensemble methods, and other models
- Concept-based explanations and methods
- User studies assessing the efficacy of explainable AI tools
- Cognitive approaches to creating effective and understandable AI explanations
- Local and global interpretation methods
- Path-based attribution, gradient-based methods, and hybrid approaches
- Mechanistic Interpretability for AI Safety
- Trust in Autonomous Systems
- Case Studies and Real-world Applications
- Policy, Regulation, and Standards
