ChatGPT in classrooms raises concerns over academic integrity and student dependence

A major international study has found that the real challenge of ChatGPT in higher education is no longer whether students use it, but how they use it. The research shows that universities are entering a new phase where the quality of student interaction with artificial intelligence is shaping learning outcomes, academic integrity, and institutional policies.

Published in Education Sciences and titled “ChatGPT at University: The Definitive Transition from Adoption to Quality of Student Interaction,” the study is based on in-depth qualitative data from 418 university students to examine how learners engage with generative AI in real academic settings. The findings point to a decisive shift in the global debate, away from simple adoption metrics toward a more complex evaluation of interaction quality, ethical regulation, and cognitive engagement.

The research highlights that ChatGPT is now embedded in the academic ecosystem, used across writing, problem-solving, tutoring, and assessment support. This widespread integration has created new tensions between efficiency and deep learning, automation and autonomy, and innovation and academic responsibility.

From tool adoption to interaction quality in higher education

The study identifies a fundamental transition in how universities must approach generative AI. Earlier research focused heavily on whether students would adopt tools like ChatGPT, often using models such as the Technology Acceptance Model or UTAUT to measure intention and usability. However, this new research argues that such approaches are now outdated, as AI has already become a routine part of academic work.

Instead, the authors frame student interaction with ChatGPT as a multidimensional academic practice. This practice involves decision-making, validation of information, ethical judgment, and communication strategies, including whether students disclose or conceal their use of AI.

Through thematic analysis, the researchers identified ten core categories shaping this interaction. These include adoption patterns, student attitudes, writing and translation practices, academic performance, cross-cutting skills, integrity, well-being, disciplinary use, and institutional integration.

The findings reveal a clear spectrum of interaction quality. On one end, high-quality interaction is marked by verification of AI outputs, rewriting and adaptation, critical thinking, and responsible authorship. On the other end, low-quality interaction involves cognitive delegation, overreliance, uncritical acceptance, and concealment of AI use.

The study shows that the impact of AI on learning is not determined by the technology itself, but by how students engage with it. In this sense, ChatGPT does not automatically improve or degrade education. Instead, it amplifies existing learning behaviors and institutional practices. The researchers also stress that interaction quality cannot be measured purely by productivity gains or faster task completion. Rather, it must be evaluated in terms of how it contributes to long-term development of critical, creative, and autonomous thinking.

Ethical tensions, academic integrity, and institutional responsibility

The study identifies a growing tension between AI use and academic integrity. As students integrate ChatGPT into their workflows, traditional definitions of authorship, originality, and plagiarism are being challenged. The research finds that low-quality interaction often includes behaviors aimed at avoiding detection or maximizing efficiency without regard for ethical standards. These practices include submitting AI-generated content without modification, failing to verify information, and concealing AI use from instructors.

On the other hand, high-quality interaction is associated with transparency, proper attribution, and active engagement with AI-generated material. Students in this category treat ChatGPT as a support tool rather than a replacement for their own thinking.

The study highlights a critical institutional dilemma. Many universities have responded to AI by focusing on detection and control mechanisms. However, the research suggests that this approach may be counterproductive, as it can encourage concealment and strategic compliance rather than genuine learning.

Instead, the authors argue for a shift toward pedagogical reform. This includes redesigning assessments, promoting ethical AI literacy, and creating clear guidelines for responsible use. Institutions that emphasize transparency and critical engagement are more likely to foster high-quality interaction.

Another key challenge is the difficulty of distinguishing between human and AI-generated text. This undermines traditional enforcement strategies and reinforces the need for structural changes in evaluation systems.

Academic integrity is not just a technical issue, but a cultural and educational one. It links student behavior to broader institutional frameworks, suggesting that universities must take an active role in shaping how AI is used.

Cognitive dependence, well-being, and the future of learning

The study explores the psychological and cognitive dimensions of AI use. It finds that ChatGPT plays a complex role in student well-being, offering both benefits and risks.

On the positive side, students report that AI tools can reduce anxiety, provide immediate feedback, and support learning in challenging tasks. These features can enhance motivation and create a sense of continuous support in academic work. However, these benefits are accompanied by emerging concerns about technological dependence. The study identifies patterns where students begin to rely on ChatGPT for validation, decision-making, and problem-solving, potentially weakening their ability to think independently.

This phenomenon is described as cognitive delegation, where responsibility for thinking is transferred from the student to the AI system. Over time, this can create an illusion of competence, where students feel confident in their work without fully understanding it.

The research also highlights risks related to cognitive overload and social isolation. As students increasingly interact with AI rather than peers or instructors, the nature of learning shifts toward a more individualized and automated experience. Importantly, the study notes that these outcomes are not inevitable. High-quality interaction involves maintaining agency, using AI strategically, and integrating it into a broader learning process. Low-quality interaction, by contrast, occurs when AI replaces rather than supports human cognition.

The findings also connect AI use to the development of transversal skills such as critical thinking, creativity, communication, and digital literacy. However, these skills do not emerge automatically. They depend on how AI is integrated into teaching practices and institutional frameworks.