HayU Blog

The blistering rise of artificial intelligence has sent ripples through all academic institutions, leaving educators scrambling to adapt. Well-founded fears about an increase in cheating, plagiarism, and other forms of academic dishonesty led to a bevy of AI detection tools such as GPTZero, Copyleaks, and Turnitin. This has led to the rise of evasion tools like Undetectable.ai and Quillbot that make AI text look more “human.” This back-and-forth laid bare the reality that we are simultaneously horrified by what AI may mean for our education system and have a ravenous appetite to keep using it. 

Much of the conversation around AI in education focuses on the potential loss of learning, writing, and critical-thinking skills. There is also a justified concern that the standards academic institutions have spent generations building will be undermined if schools and employers lose confidence that students have genuinely met the bar their diplomas represent.

The immediate response to these concerns in schools has been to develop clear AI policies, shift toward more in-class writing and assessments, and invest in AI detection tools. 

These measures all make sense. Switching to in-class, handwritten assessments eliminates the possibility of using online tools for submitted work (although it may disadvantage students with slower processing, learning differences, and writing disabilities). While investing in better AI detection tools will likely be a never-ending game of cat and mouse as AI detector evasion tools also get more robust, it seems like an unavoidable cost to bear. 

A less-discussed undercurrent in this discussion is that academic dishonesty is often linked not to malice or laziness, but to personal insecurity, a perceived lack of support, and a general lack of self-belief. Research suggests that instances of cheating decrease when: 

• Students feel like their teachers are more engaged in their academic journeys
• Students feel like their teachers are approachable
• Assessments and exams are not clustered within a few days of each other
• Students don’t feel an overwhelming pressure to perform academically
• Students have a better handle on their time management and procrastination

Incidentally, an increase in insecurity (the anxiety about oneself), even outside of the academic context, is one of the primary reasons for our reliance on AI as a society, and may point to how deep this reality runs. Recent studies suggest that AI use doesn’t lead to anxiety/depression or other mental health concerns, but existing mental health concerns often result in addictive reliance on AI. For adolescents who struggle with social anxiety and loneliness, AI chatbots can feel like safe spaces where they can increase their self-disclosure, receive advice, companionship, and what feels like empathy. The more personalized AI models become, remembering and referencing previous chats, the blurrier the line between computer and companion becomes. 

AI chatbots also seem to provide the promise to take the social tension out of the learning process. Why ask your teacher for help (a potentially difficult social interaction) and admit that you don’t understand foundational parts of the content you need to know for the test, when you can use AI to learn and eliminate the friction of that interaction? Or worse still, why spend the energy studying, researching, and drafting an essay, and potentially still not get a passing grade, when AI could potentially write it for you and save you the effort, self-doubt, and anxiety? Even though the research points to the dangers of AI dependence, and teens who overdepend on AI acknowledge that it is bad for them, they still struggle to stop. 

From a profit standpoint, the more addictive the models are (read: more sycophantic and agreeable), the better it is for the companies that create them, so it seems unlikely that this will be solved at the source. But it may be that in the academic setting, the values of differentiated learning, student-centric support, and fostering growth mindset in students will continue to be the most effective tools at countering the many risks that dot this new AI landscape. 

If students feel safe enough to try, fail, and be supported as learners, it may mean they turn to their community more often than an AI chatbot. 

Hunja Koimburi, M.A. 

Director, Hayutin Education