Why Explainable AI (XAI) Needs User Involvement
Artificial Intelligence (AI) is reshaping industries worldwide, but its complexity often leaves users in the dark. Enter Explainable AI (XAI), a concept designed to make AI systems more transparent and understandable. However, XAI itself can sometimes feel like another layer of complexity, hindering its adoption. So, why is XAI critical, and how can involving users from the start unlock its true potential?
At its core, XAI demystifies how AI systems make decisions, providing human users - especially those without technical backgrounds – with clear insights into AI's decision-making process. This is crucial in industries where trust in AI is vital for adoption. By making AI systems more understandable, XAI empowers users to become active participants in AI integration, ensuring that they not only use the technology but also trust its outputs.
However, making XAI user-friendly is no small feat. The abstract nature of AI explanations varies across contexts and user needs, adding another layer of complexity. That's why our research focuses on involving users early in the requirements elicitation process. By understanding what explanations users need - whether it’s gaining clarity, building trust, or improving satisfaction - we can tailor AI systems to better align with human goals. This approach ensures that users are equipped with the information they need to feel confident in the AI-driven decisions they encounter.
In short, XAI is more than just explaining AI - it's about making AI usable for everyone. By prioritising user involvement, we can bridge the gap between AI's power and its practical, trusted application.
Additional Resources
Author’s profile
Maria Aslam received the B.Sc. degree in computer science and the M.Sc. degree in big data science and technology from the University of Bradford, U.K., in 2012 and 2019, respectively. She is currently pursuing the Ph.D. degree with Loughborough University, U.K., with a focus on explainability in artificial intelligence and its utility and implementation in manufacturing industry.
If you would like further information on this research please email: p-ld@bath.ac.uk
