Human-Centered Explainable AI (XAI): From Algorithms to User Experiences

Explainability is not just a technical feature—it's fundamentally about human needs. The paper emphasizes that explainability has evolved from being merely a tool for data scientists to understand their own models into an essential requirement for building public trust and adoption of AI systems across various domains.

Technical explanations alone are insufficient; user experience design matters critically. Human-computer interaction (HCI) research shows that how explanations are presented and experienced by end-users is just as important as the algorithms generating them, requiring collaboration between AI developers and UX designers.

Human-centered approaches reveal hidden problems with existing explanation methods. By studying how real users interact with AI explanations, researchers can uncover pitfalls in current XAI techniques that might not be apparent from a purely technical perspective, leading to better and more trustworthy AI systems.

Different users have different explainability needs. The paper highlights that human-centered design helps identify what specific types of explanations different stakeholders actually need—whether they're domain experts, regulators, or end-users—rather than applying one-size-fits-all technical solutions.

Explainable AI requires conceptual frameworks designed for humans, not just algorithms. Beyond technical toolboxes, the field needs human-compatible frameworks that guide how to design, evaluate, and deploy explanations in ways that genuinely serve human understanding and decision-making.