It has become a common dilemma at universities whether submitted assignments were written by students themselves or generated by artificial intelligence, such as ChatGPT. The issue extends beyond possible cheating to broader questions about what we mean by "knowledge" today and how to distinguish a persuasive superficial answer from genuine understanding.
Returning to traditional methods
Some instructors are turning back to traditional exam formats: paper and pencil. For example, a university lecturer who teaches programming plans to reinstate in-class, handwritten exams next year to assess students’ coding skills without the assistance of large language models. He acknowledges, however, that this is at best a temporary measure rather than a comprehensive solution.
Pedagogical opportunities and risks
The debate about AI encompasses more than surveillance and enforcement: it asks whether a new technological paradigm can, within appropriate pedagogical frameworks, create personalized learning environments, support practice and remediation, or even foster creative work. At the same time, AI chatbots can weaken independent problem-solving because they speed access to ready-made answers instead of prompting deeper thinking.
Consequently, teachers are expected to continuously weigh in which phases of the learning process AI supports progress and in which phases it impedes it. It is not enough to vet submitted work; educators must also consider how to preserve students’ reading comprehension, critical reasoning, and cognitive abilities in an environment where producing answers has become easier than ever.
Broader ethical and historical context
The dilemma goes beyond education into ethical and historical territory: there is no single, objective standard for defining intelligence, and history shows that claims of "scientifically proven" ability have been used to strip groups of basic rights. Notably, Norbert Wiener, one of the founders of cybernetics, warned as early as 1960 that machines might eventually surpass human performance; the decisive question, he argued, would be how to ensure systems actually pursue and execute the goals set by humans rather than merely optimizing statistical approaches.
Conclusion
Using AI in education is not only a technical or enforcement issue but also a pedagogical, ethical and societal challenge. Universities and instructors must balance preserving traditional practices that develop cognitive skills with experimenting with AI-supported learning methods, while staying vigilant about when these tools facilitate and when they hinder learning.
Photo: Lecture at Széchenyi István University in Győr on 28 September 2023 (MTI / Krizsán Csaba)
