In the last year, I conducted in-person focus groups with more than 250 juniors and seniors at 20 U.S. high schools to learn how students are using AI tools for their schoolwork. I asked them which teaching strategies are more and less effective in supporting ethical, appropriate uses of AI tools. Across this mix of urban, suburban, rural, public, and private schools, the experiences students shared were remarkably consistent.
How Students Are Using AI
Students often use AI tools to navigate what they describe as busy schedules and high expectations. About 85 percent of students in the focus groups reported using AI for their schoolwork. A majority of students said they use AI to complete about 25 to 40 percent of their coursework. About 10 percent said they rarely use AI, citing ethical or environmental concerns. Another 10 percent said they use AI “all the time.”
Students shared hundreds of examples of how they use AI—to help them learn, “as a crutch,” or to cheat. Strategies for learning included doing their own work first and then asking AI for feedback or engaging the AI tool in a discussion about specific content or concepts when they were stuck. Students also shared examples of using AI to create efficiencies, such as summarizing their notes, creating study tools and practice tests, or formatting citations.
But students also shared numerous anecdotes about using AI to “take shortcuts.” These included “just putting the teacher’s questions into AI and getting the answers,” using it to complete busywork that they already understood, using it to complete complex work that they didn’t understand at all, or using it as a lifeline when they had “too much work and not enough time.”
“When teachers don’t say anything about AI, people take that as an invitation to just use it for whatever they want to use it for.”
Although students almost uniformly defined academic integrity as “honestly doing your own work,” they were quick to rationalize using AI to cope with busy schedules and maintain their grades. One student shared an anecdote that echoed what I heard in every school. After a long day of classes and extracurricular activities, and having stayed up late that night to prepare for a test, she remembered she had math homework due the next morning. “I didn’t want to take a hit in my grade,” she explained, so she used AI to complete the assignment, “knowing I would be confused in class the next day. But I just needed to sleep.”
What Students Say Doesn’t Work
Across every school, clear patterns emerged in what students say hasn’t worked to help them navigate a world where AI tools are increasingly unavoidable and powerful.
Silence as Strategy
As students pointed out, if you have access to a search bar, you have access to an AI tool that can do almost all your schoolwork. Despite this reality, students said that after the first few days of school, AI was “kind of a taboo subject” or “this unmentioned thing.” But, as one student explained, when teachers “don’t say anything about AI, people take that as an invitation to just use it for whatever they want to use it for.”
Blanket Bans
The default message that students receive from most of their teachers is some version of “just don’t use it.” But students see blanket AI bans as unrealistic. First, they know that their classmates are frequently using AI tools. Second, the messaging fails to provide guidelines around gray areas, such as embedded grammar checkers and editors, AI overview tools, autocomplete functions, or the “Would you like me to write or summarize this for you?” chatbots that greet them when they open writing software programs.
“AI Can’t Do That”
Equally frustrating for students is when teachers dismiss AI at the start of the school year by saying, “AI can’t do the work we do in this class.” Students go home that night and immediately disprove the claim. This leads students to believe that the teacher either doesn’t know how to use AI or doesn’t understand what the latest AI models can accomplish.
Too Little or Too Much Monitoring
Students notice when teacher oversight is performative or obsessive. When phone caddies are abandoned by Thanksgiving, screen-monitoring notifications are ignored, or teachers just sit behind their desks doing work during tests, many students see it as a green light to cheat. At the other end of the spectrum, when students believe a teacher is more focused on catching cheaters than teaching, honest students feel anxious about submitting their own work, and frequent AI abusers “take it as a challenge” to get away with cheating.
When students have a good relationship with their teacher, they’re less likely to cheat because ‘it almost feels like you’re betraying that teacher if you do.’
Busywork
When students are asked to do assignments that serve as “filler” or are completed and “never spoken of again,” they dismiss them as busywork. Students look for shortcuts to reserve their efforts for assignments that are interesting, relevant, and authentic. As one student explained, when “it’s packets and packets, I don’t have time to do all of this. You have to use AI.”
Enforcement Without Nuance
Students expressed concerns about the accuracy of AI detection tools, the inconsistent use of those tools among teachers, and the lengths some students go to evade detection. Rule followers can be falsely accused, and students who use AI “all the time” are always coming up with new ways to not get caught. Of greatest concern to students is enforcement without nuance, where if “a single sentence” is flagged as likely AI, “an email is sent home, and you get an F.” One senior, who was falsely accused of using AI as a sophomore, began running all her writing assignments through her own detection tool to ensure that nothing she wrote appeared AI-generated. Other students shared examples of dumbing down their writing and going out of their way to avoid high-frequency AI words or punctuation.
Rhetoric Alone
According to students, simply telling them not to use AI has little effect. Students were more likely to have their own “aha” moment about how the misuse of AI harms their learning. One student recalled a time when the teacher asked her to explain details from an assignment she’d completed by plugging the questions into AI. She realized she had no idea what any of the work she turned in actually meant. The reflection was the lesson. She noted, “I was using AI like a crutch. I was too dependent on it.”
What Students Say Does Work
In our focus groups, students also noted teacher behaviors and strategies that have helped them act with integrity, respect boundaries for appropriate AI use, and use AI tools in ways that support their learning.
Focused Lessons on What AI Can’t Do
When high school students receive hands-on lessons about AI use in each discipline, they can draw accurate conclusions about the strengths and limitations of the tool. One student shared how a teacher, in a series of lessons, had the class use AI for different academic tasks. Students then had to point out errors in the output, identify examples of bias, note the language an AI tool used to flatter the user, or catch the tool being overly confident about an inaccurate claim. These focused lessons helped students reach nuanced conclusions and apply specific strategies to use AI more responsibly.
Focused Lessons that “Teach Us How to Use AI to Learn”
Across all 20 schools, students only recalled a few specific lessons where teachers showed them how to use AI as a learning tool. Examples included a math teacher who taught students the workflow (“Do your work first, then check with AI, and then let me know what you learned”); an entrepreneurship program that offered a sequence of recurring lessons on effective prompting techniques; and an English teacher who taught students how to upload a draft and a rubric to the AI tool so it could “roast” their writing without doing the writing for them. As one student explained, if more teachers “showed us how to ask [AI] questions that help us learn, versus just giving us the answer, then a lot of students would feel more empowered to do that.”
Realistic Guidelines and Clear Boundaries
Students appreciate realistic guidelines (what they can do) and clear boundaries (what they cannot or should not do) with AI. These are most powerful when accompanied by a rationale that ties back to their learning or the assignment’s purpose. As one student said, “If one [teacher] explains when you can use it, how you can use it, and why you can use it that way, and the other just says, ‘don’t use it’ or doesn’t care,” the students who were taught the benefits and drawbacks of specific uses are more likely to listen.
Showing Them You “Get It”
When asked, “Who knows more about AI, you or your teachers?” students almost uniformly replied, “We do!” As one student explained, “I think [teachers] just don’t know how to use it themselves, which puts them in a tough spot to help us.” Students across nearly every school expressed appreciation for teachers who were transparent about how they used AI and gave them nuanced examples of how to interact with AI tools to learn rather than just get answers.
If more teachers showed us how to ask AI questions that help us learn, then a lot of students would feel more empowered to do that.
Emphasizing Learning, Not Grades and Points
Students spoke positively of teachers who have high expectations for achievement, but who don’t make everything about the grade. These teachers emphasize the importance of the learning process, of learning from mistakes and asking questions. In these classrooms, teachers reduce the misuse of AI tools because students know the teacher cares “more about what you are learning than if you just get the right answer or not.”
Minimizing Conditions Where Cheating Can Occur
Students notice when teachers go out of their way to minimize conditions that enable AI-related cheating. This included teachers who give paper-and-pencil tests or have students write initial essay drafts by hand in class so they understand that good writing is about slowing down and “actually thinking.” Students also noted that cheating is less likely to occur in classrooms where the teacher always expects them to show their work and justify their answers.
Teaching Students What You Want Them to Learn
Many students framed “good teaching” as the best way to prevent using AI to cheat. This included strategies such as providing clear unit overviews or study guides, offering hands-on learning experiences, giving students time to process what they’re supposed to learn, explaining things more than one way before moving on, and being available to patiently answer questions. One student explained that in these classrooms, “Students cheat less because they don’t have to.”
Creating a Culture of Trust
One theme recurred across every school: students said they are less likely to cheat in classes where they know it’s safe—to ask the teacher to reexplain something, to acknowledge when they’re confused, and to ask for help without feeling judged. Conversely, “If a teacher has a bad attitude, students aren’t going to ask questions or seek help. They’ll go to AI instead.” As one student explained, when students have a good relationship with their teacher, they’re less likely to cheat because “it almost feels like you’re betraying that teacher if you do.” Nodding in agreement, another student added, “And you want to try harder in that class.”
The Path Forward
As each focus group neared its end, I asked students, “What do you wish your teachers understood about AI?” One senior’s response captured the essence of what I heard in every school and across nearly every group. He matter-of-factly noted that AI has been “literally everywhere” since he was a freshman. He’s no longer surprised by what AI tools can do, nor by how much students use them to do their schoolwork. What does surprise him is how little he and his classmates have learned about how to use AI tools responsibly. “It’s going to be a part of students’ lives,” he said. “If [teachers] gave us more guidance on what to use it for, how to use it, and what not to use it for, it would be a lot better for everybody.”
