The Human-Centric AI group is passionate about developing AI technology that let people solve their problems intuitively using understandable AI systems that extend human capabilities. To reach this goal, our research focuses on explainable AI (XAI) and natural language processing (NLP).
Understanding AI systems using explainable AI
In the area of XAI, we develop methods that help us understand how AI systems work. Our methods explain the predictions of an AI system and ensure that this explanation reflects the true reasoning process of system. Likewise, we work on methods that help people understand exactly what AI systems have learned, where they might fail, and why.
Building trust through understanding
At the center of our work in XAI are human needs: we aim to enable stakeholders of an AI system to understand the system. That, in turn, enables stakeholders to collaborate with the AI system, for example, by determining if the system’s prediction can be trusted.
Advancing natural language processing
Computers are far better than humans in reading fast, storing information and organizing it. Our work in natural language processing advances computer capabilities in processing large amounts of written languages. Our goal is to improve semantic understanding of different language by computer systems to make the language processing by computers more realistic and human-like in meaning.
From knowledge graphs to semantic understanding
To accomplish this we are using information extraction techniques to transform language text into knowledge graphs that associate facts from different texts within one common representation. This allows users to browse and query large amounts of textual information more efficiently, increasing the volume of information they consume and process. Further, our goal is to explore how we can support humans in navigating multi-text information through these human-readable, highly understandable knowledge graphs.
Leading human-AI collaboration
Through both lines of research – XAI and NLP – we focus on how to build AI systems that collaborate with human users in order to support them. By combining human control with automation, we aim to create AI systems that empower people to solve complex problems, which in turn will improve society’s well-being.