According to experts, students with strong fundamentals can use AI as a learning partner by questioning its explanations, testing their reasoning and exploring alternative questions. Generative AI has entered classrooms faster than education systems have been able to establish clear rules around its use. Schools and universities are now facing a difficult choice: should AI tools be banned, detected or integrated into the learning process? Recent decisions around the world show that institutions remain divided over how to handle the technology. New York City has imposed a one-year moratorium on student-facing generative AI for public-school students from 2-K through Class 8. Meanwhile, the University of Chicago is shifting some Social Sciences Core courses to an “analog” format, where the use of AI tools, laptops and phones is prohibited. In India, Lucknow University has tightened its rules on the use of AI in PhD research. Under the revised regulations, unacknowledged AI-generated content is treated as plagiarism, with penalties depending on the extent of AI-generated or plagiarised material. Content involving 10% to 40% AI-generated material or plagiarism can be sent back for revision, while levels between 40% and 60% can lead to a one-year restriction on resubmission. If the level exceeds 60%, a student’s PhD registration can be cancelled.

These developments raise a broader question: how much dependence on AI is too much, and how should educational institutions determine the limits? Dr Ted Mockrish, Head of School at Canadian International School, Bangalore, believes neither unrestricted AI access nor a permanent ban is an ideal solution. He said schools may consider restrictions until they have clear and enforceable guidelines for students, but added that a ban should not be viewed as a permanent answer.“Banning AI doesn’t get you anywhere, nor does unfettered access. However, until schools have clear, enforceable guidelines, then banning may not be a bad option. With that said, it’s not a permanent solution,” Mockrish said. He suggested that New York City’s decision could be aimed at first limiting unrestricted access before developing practical guidelines for responsible AI use among young people. He also raised concerns about the wider risks of technology for minors, including potential child-safeguarding issues. Prof Raghuram Bharadwaj, Assistant Professor in the Department of Data Science and Artificial Intelligence at IIT Bangalore, said the focus should be on identifying what students are actually expected to learn.

“There is a fundamental difference between AI doing the thinking and AI helping communicate or execute ideas that the student has already developed. If AI is generating the arguments, analysis or reasoning that the student is expected to develop, then it is replacing the learning. But using AI for grammar, presentation or other forms of assistance is different,” he said. According to Bharadwaj, the key question for universities should be whether AI has performed the part of an assignment that the student was expected to learn independently. Karan Gupta, Co-founder of AssessPrep, believes blanket bans may not be a practical long-term solution because AI has already become deeply embedded in students’ lives. Citing a Common Sense Media survey, he said a large proportion of children aged nine to 17 are already using AI tools. “ChatGPT only went public in November 2022, and Common Sense Media’s survey this year already found 86% of children aged nine to seventeen using AI. So banning it isn’t really an option. What matters is keeping a human in the loop, where AI does the heavy lifting and the teacher still looks at the work and makes the call,” Gupta said. For India, Gupta suggested that schools avoid adopting a one-size-fits-all approach. Instead, institutions could focus on teaching AI ethics, allow teachers to decide when AI use is appropriate and require students to disclose when they have used AI in their work. Such disclosure, however, would need to be supported by systems that make AI use transparent and meaningful.