Last month, Los Angeles Unified School District announced a ban on student use of generative AI for the 2026-27 academic year, joining New York City Public Schools With the nation’s two largest school districts having led the way, other districts will soon follow.
As a recent graduate and the founder of a nonprofit working to responsibly integrate AI in education, I am all too aware of the cheating epidemic AI has fueled . Yet, I cannot help but also feel a disappointment towards these bans. Why should the institutions meant to prepare us for the world take AI—a tool which increasingly dominates the professional world, and one that has tremendously benefited my learning—away from us?
Blanket AI bans can be helpful for learning, but they are Band-aids at best. As I learned last year, sustainable solutions will require policies and environments that go beyond AI, ones that center curiosity and learning.
My university career started in 2021, meaning it was only after a year-and-a-half of banging my head against papers I didn’t want to write that AI barged into my education. Although I recently graduated, I have not left learning behind and doubt I ever will. Finding some spare time and a $20/month ChatGPT account, I set off on a quest to shore up my knowledge on the technical foundations of AI and its impacts: how did this technology come to be? What is it doing to our world?
From the beginning, AI was crucial in shaping the scope of my learning. With a desire to explore but no map, I spent an hour with ChatGPT (spent in part calling out hallucinated sources), and found myself with a syllabus personalized to my goals and knowledge.
Instead of a teacher, I had ChatGPT as a companion. Just as I might turn to a friend or go to office hours, I turned to AI to dig deeper than a reading went and clear up uncertainties. This was not only faster, but I was also more free to ask questions I might otherwise think silly. I could learn as fast—or more often than not—as slow as I wanted.
As I worked through content and my understanding grew, it was undeniable that AI’s impacts were beneficial, fittingly, on my understanding of AI’s impact on society. Questions that might have once required hours of digging on my own were answered in minutes, and where I found myself wanting to dig deeper into a topic or starting to yawn, I only had to ask ChatGPT to redesign the syllabus to fit my needs.
At the end of a month and a half, I had worked through four books, 20 hours of lectures, dozens of articles, and written four essays reflecting on what I had learned. The majority of this time was spent engaging with existing content, and there were also times when I found AI steering my astray. Yet on the whole, AI had undoubtedly made the process smoother, faster, and more productive than if I had approached it on my own.
Learning with AI made me smarter. So why is it having the opposite effect in schools?
Evaluating the impact of AI on learning requires a closer examination of the process of learning. On the one hand, learning requires content. In this regard, LLMs follow a rich tradition of technologies like the TV, smart phone, and MOOC that excel at delivering information. There is very little a student today cannot learn through a combination of Youtube videos, online forums, and conversations with ChatGPT.
On the other hand, learning is not a productive act, but an experiential one. AI personalized my learning process, but it could not do the learning for me: learning requires desire. For a student motivated to learn, AI can enhance that thirst of knowledge. But for a student primarily motivated to complete assignments, AI enables cheating and cognitive dependencies. The same technology in different settings produces opposite outcomes.
To prepare students for a world with AI, then, bans are not solutions. Instead, we need to create environments that nurture student’s innate human desire to learn. We need teachers who have done what good teachers have always done—kindle the spark in their students’ eyes, challenge students to grow, and stand up for what’s right—and honest conversations about the structural incentives that drive students to cheat rather than learn.
Until then, we are left applying Band-aids that, in the name of learning, prohibit students from using the world’s best learning tool to learn.