Lecture: Communicative AI in Schools at the Open Campus 2026

The Use of Communicative AI: Between Educational Opportunities and Risks for Learning

Bremen, June 13, 2026 – At the University of Bremen’s Open Campus 2026, Prof. Dr. Karsten D. Wolf from the Lab Media & Education at the ZeMKI (Center for Media, Communication and Information Research) presented a comprehensive scientific analysis of the impact of generative artificial intelligence on the education system. His lecture addresses the pressing question of whether AI should be understood primarily as an educational opportunity or whether it represents a fundamental threat to learning.

The Status Quo: AI Has Arrived in Everyday Learning

The relevance of the topic is reflected in the rapid adoption of the technology by school students. According to data from the 2025 JIM study, 50% of the young people surveyed now use ChatGPT daily or several times a week. The use of AI applications for homework or for learning rose significantly from 65% in 2024 to 74% in 2025. AI systems, which—as probabilistic text production systems based on vast amounts of data—learn language patterns and make predictions, have thus become a firmly established part of learning culture.

Educational Opportunities: The Teacher of the Future and the AI Agent Swarm

Prof. Wolf makes clear that AI offers enormous didactic potential to support teachers. The technology enables deeper individualization and differentiation in the classroom. With detailed prompts and the provision of background information (context engineering), tailored learning materials can be generated for different performance levels.

The current vision for the “teacher of the future” in the research context envisions a synergistic model: supported by an “AI agent swarm,” AI helps to analyze data, carry out real-time differentiation, and deliver immediate feedback. This relieves teachers of administrative and diagnostic routine tasks, allowing them to focus more strongly on relationship work, mentoring, and the development of social competencies.

Risks for Knowledge Acquisition: “Cognitive Debt” and the Unlearning of Skills

These immense opportunities are matched by equally serious empirical risks. Among the primary dangers are the unlearning of basic skills, a possible erosion of intrinsic motivation, the lack of transparency in these systems, and the inherent biases of the models.

Particularly critical is the phenomenon of so-called “cognitive debt” when writing essays using LLMs. A cited study shows this strikingly: participants who used AI assistants to compose texts subsequently had immense difficulty (over 80%) quoting or reproducing content from their own essays—in stark contrast to control groups that used classic search engines or no aids at all. This underscores the danger of outsourcing cognitive processes and actual thinking to the machine. Further challenges include technical unreliability (hallucinations) as well as excessive dependence (overreliance) on the technology.

Conclusion: “Teaching- and Learning-Related AI Literacy” as a New Core Competency

To avoid the so-called Matthew effect—in which already advantaged learners with good learning strategies benefit disproportionately—the integration of technology must be actively steered in pedagogical terms. This, however, also requires a specific teaching competency relating to dynamically evolving AI technologies. Prof. Wolf concludes that building a specifically teaching- and learning-process-related AI competency (AI literacy) and specific learning strategies, as an integral part of media literacy for both teachers and learners, is indispensable. Even in a learning environment permeated by AI, one thing remains true: genuine educational processes still need human beings.