Illustration by Celine Bejjani
My biology professor encouraged us to use an AI-driven platform called Connect to help us prepare for our exams. So I did, almost blindly. I went through the questions, memorized the automated answers, and poured my full trust into the system. But when the results came out, my grade was unexpectedly low.
I decided to sit a second exam. This time, I read the entire textbook and cross-referenced with external sources. My results? Much better. The difference was not simply in the grade, but in the depth of understanding. When I relied on AI, I memorized patterns of answers. When I relied on the textbook, I understood the logic behind them. The platform trained me to recognize what sounded correct, but not to grasp why. In hindsight, I realized I had outsourced not only my preparation, but indeed the very thinking process.
The experience forced me to confront a more unsettling question: what happens to learning when thinking itself becomes automated?
Which compelled me to ask another set of questions: Does the use of AI in educational settings genuinely support learning? How does it shape students’ thinking and understanding of knowledge? Does it limit freedom, spontaneity and meaning?
As the use of AI in education becomes increasingly common, so is the effect it has on traditional methods of acquiring knowledge. The increasing integration of AI into classrooms risks transforming education from a space of intellectual exploration into a system of algorithmic compliance, subtly reshaping how students think, question, and exercise freedom [4]. AI systems generate probabilistic answers, not verified truths. They predict coherence, not correctness.
Writing should be a tool for processing and expressing one’s thoughts and learning. However, with AI-driven learning, writing tends to be reduced to a basic task rather than an intellectual engagement. The more AI adapts to its users, the more the answers it produces sound complete, emanating a level of absolutism that is hard to refute.
Cedric John, professor and Head of Data Science for the Environment and Sustainability at Queen Mary University of London, points out that one of the risks with AI-generated content is that it can appear convincing while still being unreliable. It often “lacks objectivity” and can even “cite work that is sometimes inappropriate or even imaginary.”
These traits become stronger over time, with students increasingly treating their answers as final, having lost the skills to question, think critically, or seek alternative perspectives. Where face-to-face interaction between students and their professors used to be the main way to ask questions, seek clarifications, or engage in collective, even heated, intellectual debate, students today are opting to turn to AI instead of their peers or instructors when seeking answers. Instructors, too, are starting to use AI to generate and grade assignments.
Professors, too, are tempted to follow charts, predictions, and the scores they receive from these automated systems instead of engaging with an organically generated pool of ideas. With time, learning becomes less about intellectual curiosity and more about ticking a series of boxes in order to pass to the next level.
However, AI is not entirely inherently detrimental to education. It can democratize access to information, assist students with different learning needs, and provide instant feedback that traditional classrooms often cannot offer. In many contexts, it enhances efficiency and allows learners to explore topics beyond the limits of their textbooks. The problem, therefore, is not the existence of AI but the uncritical dependence on it.
When AI becomes the primary source of answers rather than a supplementary tool, it subtly reshapes intellectual habits. Building on this and other concerns, the appropriate response is not to reject AI outright, but rather to change how it is used within educational settings. Instructors should encourage class discussions and integrate them as an essential part of the curriculum. Assigning peer-work activities can also promote student interaction.
At the same time, students need to be better guided on how to think critically about AI outputs. Since AI can produce information that sounds confident, yet is inaccurate, teachers must help students identify errors and evaluate sources. As John (2025) observes, AI tools function like a “know-it-all bot always willing to help you with your work,” which makes them highly accessible but also potentially misleading if their responses are not carefully examined.
Assessments also need to be redesigned to ensure they evaluate reasoning and understanding. This can be done through in-class tasks, reflections and projects that show students’ thinking processes.
Many teachers and students currently receive little to no training on how these systems work or how to monitor them responsibly. By giving teachers and students clear instructions, simple rules and proper training, schools can make sure AI does not become a tool of thought control.
Similarly, when education is reduced to automated feedback and machine-generated certainty, students are no longer encouraged to think, but to comply. This erosion of human thinking contributes to the total domination that Arendt describes. As she argues, the loss of thinking is the precondition for domination. Therefore, preserving meaningful education requires ensuring that AI remains a tool that supports humans, not replacing them.
As the Cheshire Academy suggests, students may use AI for “brainstorming and introductory research,” while other parts should be done on their own.
Banning something as prolific as AI is neither realistic nor necessary. The real danger lies in letting it turn education into a system of automated certainty rather than intellectual inquiry. When students stop questioning and begin accepting algorithmic answers as final, thinking itself is weakened. AI must remain a tool that supports human judgment and not a system that replaces it.
