Transferring Domain Knowledge with (X)AI-Based Learning Systems

XAI can serve as a cost-effective teaching tool: The study demonstrates that explainable AI systems trained on expert decisions can successfully teach novices complex tasks, offering a scalable alternative to expensive one-on-one expert training in high-stakes domains.

Explanations matter for learning outcomes: The research shows that providing AI-generated explanations alongside examples significantly improves novices' ability to learn classification tasks, suggesting that transparency in AI systems has direct pedagogical value beyond just interpretability.

Individual differences affect learning from AI: The study of 249 participants found that learners' cognitive styles moderate how effectively they learn from (X)AI systems, indicating that one-size-fits-all AI tutoring may be suboptimal and personalization could improve educational outcomes.

XAI has untapped potential in human-AI collaboration: This work represents early evidence that explainable AI can be deliberately designed for human learning purposes, opening new possibilities for AI developers to create educational tools rather than just interpretable decision-support systems.

Tacit knowledge transfer is achievable through AI: The research addresses a real challenge in professional training—transferring experts' intuitive, hands-on knowledge to novices—by showing that AI systems can capture and communicate this expertise through structured examples and explanations.