NEWS A new field of research, known as Social Explainable AI, is taking shape internationally within AI. It focuses on making the decisions taken by AI models more understandable to the general public. “Current explanations are primarily aimed at experts and do not meet the EU’s transparency requirements under the new AI Act. We want to change that,” says Kary Främling, whose new book has attracted international attention.
The new book Social Explainable AI brings together ideas from many different fields, including AI research, linguistics, philosophy, psychology and the social sciences. "It places users, context and communication at the centre,” Prof. Kary Främling, explains.
ImageMattias PetterssonKary Främling, professor at the Department of Computing Science at Umeå University, is one of the most cited researchers in his field. His work in Explainable AI, the Internet of Things, and digital twins has laid the foundations for the systems we use across the globe today. Together with international researchers, he has highlighted the gap that has emerged between the tools and systems available to explain how AI models used by business and public authorities reason – and how they make decisions.
"We are seeing that the explanatory models are primarily aimed at professionals, who use their knowledge to gain insight into AI’s black box. But those affected by government decisions or whose loan applications are rejected do not have the means to absorb and understand the information,” says Kary Främling.
Together with Prof. Dr. Katharina Rohlfing at the University of Paderborn, and Brian Lim, Associate Professor at the National University of Singapore, he has brought together a large number of researchers from various fields to develop a shared vision. The result is the new book Social Explainable AI, a 30-chapter handbook that has attracted attention, particularly in Germany.
The book brings together insights from AI research, linguistics, philosophy, psychology, and the social sciences to re-examine the concept of explainability as a social, interactive process.
“Instead of focusing solely on AI models, it places users, context and communication at the centre,” Kary Främling explains.
The EU AI Act requires transparency – ensuring that everyday users have insight into decisions that affect their lives, such as why they are denied a loan, fail to secure a job, or how they can increase the value of their home.
Within this new field of research, experts aim to ensure that AI models genuinely tailor their explanations to the recipient’s level of knowledge and needs.
"We want the models to use language and a level of detail that is understandable and manageable. That the information is provided in appropriate steps, and that users should be able to interact and ask questions about more specific details." Kary Främling gives an example:
"If an AI model values a detached house at five million, it should first explain the decision using understandable facts such as condition, size and location, rather than, as is currently the case, providing the enquirer with a list of numerous of technical parameters."
"End-users should have greater insight into decisions that affect them, which is something that is currently in high demand and attracts considerable interest. We hope that those working in the field of Explainable AI will take note of the book and develop their methods in a direction that takes better account of end-users, says Prof. Främling. "But also that end-users, companies and other organisations realise that they can, and should demand more from AI systems in terms of explanations in order to comply with the AI Act."
The Department of Computing Science has expanded at a record pace in recent years. Several international researchers specialising in areas such as AI, privacy, data security and autonomous distributed systems are now based here in Umeå. Kary Främling is one of those affiliated with WASP – the Wallenberg AI, Autonomous Systems and Software Programme, Sweden’s single largest research programme.
The courses offered by the department are closely linked to research, and Kary Främling teaches the course "Data Preprocessing and Visualisation". Here, students learn to analyse and present data and statistics in an accurate and humanly understandable way, so that decisions and actions can be taken based on rational information.
In 2025, he also launched a new course on Explainable AI, which covers fundamental principles, methods, challenges and even sXAI.
“Data science is one of the most important fields when it comes to shaping our current and future society. Regardless of sector or scale, organisations that want to remain competitive must keep up with developments or risk falling behind,” says Kary Främling.
Read more about Professor Kary Främling and visit the research group web page, XAI, Explainable AI. Read the book, Social Explainable AI here, (open access).
