The AI Conversation We Need to Be Having With High School Students

Kristi Kraychy, BMus, BEd, MEd.

September 12, 2026

At the beginning of both this year and last, we talked about AI: we discussed academic integrity, we have policies and we clearly explain expectations. All of that matters, but you can’t stop at the foundation and call the house finished.

It feels a little like telling a toddler before entering a store, “we aren’t getting a cookie today”, then walking them past the most delicious display of cookies imaginable, perfectly positioned at eye level… we probably shouldn’t be surprised when they reach for one!

For teenagers, AI is right there. It’s fast, polished and reassuring. High school students, especially those striving for post-secondary schools, also desperately want and need good marks. They’re afraid of being wrong, afraid of submitting something imperfect and are increasingly starting to believe their own writing is not good enough without AI making it sound better.

I am the principal of an online and hybrid school and we don’t have the luxury of returning to paper and pencil or banning technology while we figure it all out. So, my team and I are constantly experimenting, collaborating and trying new things. 

As is so often the case in education, and really in all human interactions, strong relationships are still the most important part of the solution. 

At the start of a course, as always, we have been building relationships with students first. Then, we have been calling out poor AI use quickly - not in an accusatory way - but using both empathy and directness in tandem. We have found the key is to have a dialogue in-the-moment (or as soon as possible). We are finding that, this year, the academic integrity lectures at the start of the year are not really landing for them.

While we still enforce consequences such as re-doing the assignment because it can lead to important learning, we have not been jumping immediately to extreme disciplinary action or citing the student code of conduct. We just don’t think that is how start-of-school-year relationships should begin. I’m also not convinced that inappropriate AI use is really about academic integrity at this stage in its existence anyway… at least not yet.

These tools are incredibly new, remarkably powerful, and suddenly available to all of us. It would be unfair to pretend that adults have figured out how to navigate them with integrity either. In many ways, we are all still toddlers in the cookie store when it comes to AI.

My greater concern is how quickly students have lost confidence in the value of their own minds.

When an AI tool can instantly write something more polished, solve a problem faster, or generate an idea that they think sounds better than their own, students can start to assume that their thinking is somehow less valuable.

Strong students are second-guessing their own work, don’t understand what we mean when we say “use your own voice”, and look to AI for reassurance before they trust themselves to submit even a rough draft. 

Before we reduce this to a conversation about cheating or even the more progressive conversation about outsourcing the hard work of learning, we need to recognize that students are already losing trust in themselves as learners and creators. 

So, I firmly believe our job as educators is to educate, build that trust and boost confidence first, rather than punish.

Until we are given a curriculum that explicitly and expertly teaches students how to use AI tools well and responsibly, how to know when those tools are supporting rather than replacing their thinking, and how to recognize the difference between machine-created and authentically human work, we are going to have to assume less, teach more and figure some of this out in real time alongside our students.

We are only a few weeks into the school year, and I have already seen this play out almost identically in different online courses.

Our high school students were given a variety of assignments in different subjects that had been intentionally and thoughtfully re-designed for the age of AI. These assignments avoided simply regurgitating facts. We were asking students to be creative in numerous ways and make personal connections between the course content and their own lives or worldviews. 

Many still used AI.

The bulk of student submissions were polished, generic and strikingly similar to one another. More importantly, despite the rubric and assignment outlines, the finished product told us almost nothing about the students who had submitted them.

We called the students in and were direct and firm, showing them the AI “tells” that were problematic and obvious. But rather than a long lecture on academic integrity, we asked questions about the “why” with curiosity and compassion. This dialogue included some tears and several students were clearly embarrassed. Yet because we started with the relationship first, there was also a lot of honesty, vulnerability and growth. 

Students told us they were worried their own work and writing wouldn’t be good enough. They talked about the pressure to get high grades, their fears about university, and wanting to be sure that they were on the right track. Some described turning to AI to help them get started with ideas or just at the end to “check” that it met requirements. Somewhere along the way, though, the help had begun to replace more of their own thinking than they realized.

So we reminded them that we weren’t asking for perfection, especially at the start of a new course and school year. We explained that work can and should be a bit messy while they are developing their ideas, learning new things or figuring out their unique style and voice. We explained that this is how we can give useful feedback and help them get from point A to point B. We also told them that we find their own imperfect work far more interesting to mark than polished, generic machine replicas.

We framed next steps more as an opportunity: take the low mark because the work had not met the requirements of the rubric, or re-submit the assignment without using AI. We also promised that later in the course, there would be opportunities to improve their grade by demonstrating their growth and true learning over the semester.

Our goal now is for students to see value in their own work, ideas and authentic voice, and to find reward in the struggle and growth that comes with real learning. Their final mark should reflect that hard work and the skills they have developed by the end of the course, rather than punish them for not being perfect from the start.

This also means re-thinking some of our traditional approaches to cumulative assessment in high school, so that final grades better reflect what students know and can do by the end of a course, rather than how well they performed while they were still in the early learning process.

There will be more conversations, more teaching, and we will be more clear with our expectations for each assignment. There will also be times when disciplinary action is appropriate and necessary, and that too, can be an important part of learning.

But I am starting to see a bigger problem emerge than students “taking the easy way”, “cheating”, or even the tendency to avoid some of the struggle necessary to grow and learn: they are beginning to trust AI more than they trust themselves. And I think we need to pay very close attention to that.