Enhancing AI Explainability Through the EXACT Framework: A User-Centric Approach

The increasing adoption of Artificial Intelligence (AI) in several industries has created a demand for user-centered explanations that align with how users think and understand concepts. This paper presents EXACT (EXplainable AI with Cognitive Theories), a novel framework that combines cognitive theories that explain howpeople think and understand with cognitive functions, focusing on perception, memory and language abilities, to improve users’ comprehension of and engagement with artificial intelligence technologies. By aligning cognitive functions with the design principles of Human–Computer Interaction (HCI), which promote user-centered intuitive systems, the framework addresses challenges related to making AI understandable to users with various levels of cognitive abilities. As a proof-of-concept, a self-diagnosis tool was created to demonstrate the framework’s effectiveness. Then, 60 participants were divided into a control group and an experimental group. Participants completed six tasks designed to evaluate their perception, memory, and language-related cognitive functions. The experimental group outperformed the control group across all tasks, demonstrating significantly improved performance. Subjective metrics also supported these findings: the experimental group reported higher levels of understanding (4.60 vs. 2.87), confidence (4.67 vs. 3.07), and clarity (4.87 vs. 2.80) compared to the control group. These findings suggest that EXACT framework significantly enhances user’s functions when using AI systems. However, further research is needed to explore its broader applicability in other contexts and utilize other cognitive functions.