Empirical evidence on opportunities and challenges of generative AI in education
Stefan Küchemann
Since the development of ChatGPT, the opportunities of generative artificial intelligence (GAI) in STEM education have been massively increased. The opportunities range from the automatic generation of virtual experiments, a multistep data analysis of research data, over the code generation for programming tasks to the feedback of students during learning. Besides these substantial possibilities, there are several concerns and challenges that arise from data protection to incorrect output, misuse and users’ cognitive decline. In this talk, we present an overview of empirical evidence on opportunities and challenges that arise from the integration of GAI into higher education in STEM. Specifically, we focus on how to effectively collaborate with the GAI during learning and teaching and highlight sensitive aspects that lead to more effective learning rather than a cognitive overload or decline during the use of GAI.