AI is Disrupting Education – For Better or Worse

"Empirical data on writing style show that large language models are changing our information practices more than the Covid pandemic," emphasizes Anselm Küsters, cep digitalisation expert and author of the study. AI-generated content is already influencing many scientific processes, for example by complicating the usual peer review or systematically manipulating scientific journals. There is also a growing digital divide within scientific disciplines based on a lack of knowledge about AI, which is hindering scientific progress. Furthermore, there are structural concerns about the role of AI providers in education, such as asymmetric access to state-of-the-art AI services and significant development and implementation issues with non-English language models.

"Providing every student with a ChatGPT account is not enough to improve learning," Küsters argues. Early experiments have shown that careful adaptation of AI tools is essential for effective integration into education in order to achieve a lasting effect. To enable responsible use of the new technology, AI courses should be mandatory at all levels of education and cover not only the technical basics, but also ethics, practical applications, and general problem-solving skills. Targeted support for language models based on under-represented languages is also needed, as well as specific internal guidelines to monitor the ethical use of AI in education and ensure transparency. "On this basis, generative AI can foster creativity and make knowledge more accessible, in addition to the data analysis and writing support that is already common. This is essential for Europe's future competitiveness," concludes Küsters.