DESCRIPTION
Artificial intelligence holds significant promise for streamlining routine and repetitive tasks, yet many AI projects still struggle to succeed in real-world settings. Research points to a lack of user involvement as a key factor behind these challenges. However, conducting effective user requirement analysis in AI is notably more complex than with other technologies. Challenges include unrealistic user expectations, difficulties in model explainability, rapid and unpredictable advances in AI, and the intricate social contexts into which AI is deployed. Design thinking offers a structured yet flexible approach to address these challenges by enabling teams to deeply understand users, question assumptions, redefine problems, and develop innovative, testable solutions. Since its origins in the 1950s and 60s, design thinking has proven highly effective in product innovation, stakeholder engagement, and solving complex, ambiguous problems. This tutorial introduces the five phases of design thinking—Empathize, Define, Ideate, Prototype, and Test—illustrating why and when to use this approach in AI research and development. Participants will explore specific frameworks and guidelines for integrating design thinking into AI projects, supported by case studies demonstrating its application in AI contexts. The session will culminate in a hands-on workshop focused on improving the selected principles of dataset quality through design thinking principles. Results from the workshop will be shared with AIoD and other European AI platforms, contributing practical insights to the broader AI community.
DETAILS
Course type: Tutorial (in person delivery)
Duration: 1 hour for the lecture, 1 hour for the workshop
Institution of the lecturer: Kempelen Institute of Intelligent Technologies
Notes: Presentation of the findings from design thinking workshop. Presentation will be shared by all members of the groups.
LECTURER
Andrea holds a PhD in Library and Information Science. Her background is useful in human-centered and data-driven aspects of artificial intelligence design. Andrea developed a passion for researching user needs and customizing AI tools to meet them. She believes that AI can support a wide range of users, from fact-checkers to doctoral students in AI. Andrea’s research interests also include exploring qualitative aspects of datasets for AI models. She has contributed to expert dataset preparation, including datasets on gender stereotypes and disinformation, and has participated in various experiments leveraging these datasets.