AiTech Agora: Kostas Tsiakas - Designing Human-AI interactions using Explainable and Human-in-the-Loop AI
08 June 2022 12:30 till 14:00 | Add to my calendar
In this talk, I will briefly present my work on Human-AI Interaction and the applications of Machine Learning for user modelling and personalization of human-centered AI (HCAI) systems for applications in healthcare and education. More specifically, I will discuss different cases of human-AI interactions with respect to human autonomy and user involvement in the learning and decision making process. I will present my insights from three different HCAI prototypes (socially assistive robots, computer vision for cognitive assessment and personalized game-based learning), considering the levels of interaction and autonomy between different types of human users and the algorithm of an HCAI system. Taking into consideration existing frameworks and guidelines for designing human-AI interactions, the goal of this talk is to highlight the potential benefits of integrating human-in-the-loop (HITL) and Explainable AI (XAI) methods into the design process of human-centered AI systems, and discuss future directions towards a new design space and design possibilities for collaborative learning and hybrid intelligence systems.
Konstantinos (Kostas) Tsiakas is a postdoctoral researcher at Human-Centered Design since January 2022. His research interests and experience lie in the intersection of Artificial Intelligence, Machine Learning and Human Computer Interaction, focusing on the design and implementation of personalized human-centered AI for applications in education, healthcare and future work practices. He is interested in designing AI prototypes for HCI in order to investigate: (a) how to improve AI methods for HCI (e.g., user modelling, personalization), and (b) how to improve user experience by integrating human-AI interactions (explainable and interactive AI) in the design process, focusing on aspects of human perception, autonomy, and performance.
In January 2020, he joined the Industrial Design department at TU/e, as a postdoctoral researcher working on Explainable AI and User Profiling methods to support self-regulated learning in students. In 2018, he joined the Yale Department of Psychiatry as a postdoctoral researcher, working on computational methods (machine learning and computer vision) to assess children’s executive functioning through body movements (embodied cognition). He received his Ph.D. in Computer Science and Engineering in 2018, from the University of Texas at Arlington. His PhD research focused on the design and implementation of personalization methods for social assistive robots and cognitive training.
This hybrid event is a part of the AiTech Agora meeting series. The meeting will take place physically at TU Delft, yet it is also possible to join it online. If you want to join this meeting either in person or online, please email aitech@tudelft.nl. To automatically receive biweekly invitations with invitations for the future meetings, sign up for our mailing list.