Critical Thinking Crisis: How AI is Homogenising University Students' Thought Processes

In the seminar rooms of Yale University, Amanda (a pseudonym) began to notice something amiss. Her classmates sat behind laptops with seemingly perfect argument points, yet class discussions often felt bland and flat. “Everyone sounds the same now,” Amanda told CNN. She once caught a classmate frantically typing, asking the professor’s question directly to a chatbot for an instant answer. This phenomenon marks a major shift in the world of education: AI is not only assisting with tasks but beginning to erode humans’ ability to think authentically. A study published in the journal Trends in Cognitive Sciences in March confirmed Amanda’s concerns. The research found that large language models (LLMs) systematically homogenise human expression and thinking in three dimensions: language, perspective, and reasoning. Because AI is trained on data dominated by Western, educated, industrialised, rich, and democratic (WEIRD) perspectives, the responses generated tend to reflect a narrow slice of human experience. As a result, the diversity of ideas in class discussions—where innovation should be born—is starting to flatten. Jessica, another senior student at Yale, admits to using AI every day. She even uses chatbot assistance just to formulate sentences to sound more cohesive. However, she acknowledges its negative impact. “My work ethic has really decreased compared to when I was in high school. I feel lazier,” she revealed. Thomas Chatterton Williams, a humanities professor at Bard College, sees the paradox here. While AI “raises the standard” of class discussions to be generally better, this technology is killing off eccentric and original thinking. “My biggest concern is that many smart young people will never find their own voice, and they won’t even fully appreciate the value of owning a perspective,” Williams emphasised. Facing this challenge, lecturers at Yale and other universities are beginning to change their teaching strategies. Sun-Joo Shin, a philosophy professor at Yale, now reduces the weight of homework grades and focuses more on oral exams and class presentations. Some educators are even applying “traditional” methods to maintain academic integrity: Morteza Dehghani, a professor of psychology and computer science at the University of Southern California, warns that if humans continue to hand over the reasoning process to AI, society will lose the ability to critique mainstream ideas or political candidates. AI should be positioned as a collaborator to find weaknesses in our arguments, not an agent that does everything on our behalf. Without clear boundaries, future generations risk graduating without ever experiencing the “creative suffering” of constructing ideas, a process that is the core of learning to think.