“How do we stop students from cheating with AI?” It’s a question that has been repeated numerous times since generative artificial intelligence has shown up in the classroom. It’s an honest concern when AI can write an essay, solve a problem in seconds, or summarize a chapter. The real question that I have focused on through my doctoral research and working with students is how we teach students to use AI. Specifically, how can we have them use it while helping them think critically, grow their language, and strengthen their human connection; everything that a school should incorporate.
One researcher studied K-12 grade teachers in Idaho and found that most weren’t ready to integrate AI into their teaching. Most of these educators leaned toward lesson planning or grading rather than actual instruction; less than half used it at all. Many of the concerns for not using it were due to privacy, bias, ethics, and a lack of school policy. They also often worried about less human interaction in the classroom (Cheah, 2025).
Azzam and Charles (2025) found a similar pattern in their interviews with STEM teachers in the UAE. These teachers were enthusiastic about the possibility that AI might personalize lessons, break down complex concepts, and lighten the workload. However, they did bring up concerns about inaccurate answers, data privacy, and the risk to students’ critical thinking. It’s important to note that researchers noticed that optimism didn’t translate into good pedagogy.
A systematic review by Kashif et al. (2025) pulled these threads together across the wider research base. They argue that schools shouldn’t just decide to adopt AI and think they are finished. They need to ask harder questions about equity, sustainability, appropriateness, and whether the approach can actually scale. Access matters. Ethics matter. Fit with instructional goals matters. None of that happens automatically.
So the question schools need to address is whether AI is opening a door to learning or quietly closing it. If a student uses AI to understand one tricky word and then goes back to the original text, they’re still reading, still interpreting, still doing the work. If AI summarizes a whole text and writes the response, did the learning actually happen? This distinction matters even more for students who are still developing their English language acquisition. A flawlessly polished AI-generated paragraph can hide exactly what a student understands and where they’re actually stuck. A teacher can’t truly support the needs of their students if they can’t see their academic language and/or thinking.
At Besant Hill School, this is part of the conversation we are continuing to have as a learning community. With students bringing different academic, cultural, and language experiences into the classroom, AI can be a useful support, but it cannot replace the thinking, discussion, collaboration, and relationships that are central to learning. Our goal is not simply to tell students when they can or cannot use AI. It is to help them understand when it supports their learning and when it begins to do the learning for them.
When considering AI, an important question to ask is: Did it clarify something? Did it generate the whole answer? Did it support language? Did it support curiosity, or did all the students skip the practice? Could a student explain, revise, and/or defend their work without AI?
Students should be taught AI literacy, the ability to understand, apply, evaluate, create, and navigate AI ethically (Cheah et al., 2025). None of this is about centering AI in education. Moreover, it’s about preparing students to make good technology decisions. As students encounter AI in school and beyond, it helps to guide them with practice. At the same time, teachers need real, sustained professional learning and support for families with expectations. To foster a strong learning community, it relies on what we skillfully ask students to do for themselves.
