Human AI
Metacognition in human-AI interaction.
Period:
06.2023 – 06.2025
Prof. Dr. Monika Undorf | FB 3, Angewandte Kognitionspsychologie
Prof. Dr. Ekaterina Jussupow | FB 1, Wirtschaftsinformatik
---
Project description:
People are increasingly working on complex cognitive tasks with systems based on artificial intelligence (AI). AI?supported decision-making presents new challenges for human decision-makers: in addition to their own problem-solving skills, they must also monitor, interpret, evaluate, and refine the AI's suggestions. Human metacognition is crucial for these activities. In this project funded by the Forum for Interdisciplinary Research, we combine methodological and theoretical approaches from psychology and business informatics to investigate the role of metacognition in AI-supported decision?making. Our results show that high metacognitive abilities are an important success factor for AI-supported diagnostic decisions made by medical experts. On the other hand, students rate solutions they have developed together with generative AI as just as correct metacognitively as solutions they have developed without AI support. This enables students to assess very well how well the AI system helps them. The results of the project have been presented at international and national conferences, have been incorporated into three third-party funding applications, and form the basis for two submitted and four planned scientific publications. During the project, topics for around 20 theses and for scientific networking within and outside TU Darmstadt emerged. The funded project also lays the foundation for a more comprehensive interdisciplinary research program on metacognition in human-AI collaboration.
