High schoolers now need an AI driver's license too

A four-part framework to make students drivers, not passengers

My first car was a used, white 1997 Toyota Camry with a spoiler and a six-disc CD changer in the trunk.

This is not a photo of that car. If I ever had any photos of it, they were the old-fashioned kind that had to be developed and so could be lost.

Back when photos could still be lost, there was only one kind of driver’s license high schoolers needed.

Now there are two.

As with a traditional driver’s license, saying high schoolers need an AI driver’s license isn’t the same thing as saying they need to drive. We want young people to know how to drive so they can if they choose to.

We’ve accepted cars as part of our landscape and our lives, in the US in particular. (An increase in train travel, mass transit, and bike riding would of course be ideal as far as the climate is concerned, but that’s a subject for another Substack.) As a result, though we don’t legally mandate driver’s licenses, we do culturally consider them necessary.

For good reason: cars, once a revolutionary new technology in their own right, can help us get where we want to go faster and more easily. Because of their unique affordances, we’ve in many ways built our physical lives around them.

So we think of the ability to drive a car as a necessary skill. That’s the case even as we acknowledge that cars’ benefits come with many costs (safety and pollution chief among them). And we think of it as a skill everyone should have, whether or not they choose to drive most days — because there are times when driving a car has become almost unavoidable.

To understand the analogy fully here — and the corresponding need for an AI driver’s license — we need to go back from my ‘97 Camry to the ‘15 Model T.

Six-disc changers did not come standard.

AI in 2025 is Cars in 1915

ChatGPT’s launch in 2022 heralded the arrival of the Gen AI era.

The Model T’s launch in 1908 did the same for mass-affordable automobiles.

In 1915, the one millionth Model T rolled off the line.

In 2025, it feels like more than a million AI tools and products have rolled into our lives.

3 years since its Model T moment, ChatGPT is used by 700 million people every week. Or as Ethan Mollick puts it, “More than a billion people use AI chatbots regularly.” And it can feel like the underlying tech is accelerating even faster than its adoption.

It’s as if, in car terms, we’ve jumped from the 1915 Model T to the 1997 Camry in 3 years. But, in terms of social and political infrastructure, it’s like we’re driving late 20th century cars on early 20th century unpaved roads, with spotty local regulations and no national highways.

It took until the 1950s for driver's licenses to become a nationwide commonplace. We can’t wait until the 2050s — or even the 2030s — for an AI driver’s license framework to become a commonplace pedagogy. Students need to get onto the road so they can get into the conversation and help shape the world they’re graduating into.

An AI Driver’s License Framework

I came to this idea after creating and teaching AI-related lesson plans since 2022, co-designing and piloting an AI literacy curriculum last year as a fellow at Stanford Digital Education, and running AI bootcamps for students on two continents this past summer.

I also came to it after 21 years of teaching, the last dozen teaching 12th grade. So for 12 years I’ve seen students run into class the day they receive their driving permits, waving them in the air like so many Charlie Buckets who have at long last found their golden tickets.

High schoolers want to learn to drive. That process is deeply ingrained in our culture, and it’s obvious to students why they need to go through it. They also go into it knowing (at least in a theoretical, teenage way) that driving involves profound questions of responsibility and safety. They understand the promise and peril of getting behind the wheel.

For all those reasons, they’ll understand the implications of an AI Driver’s License too, the promise and peril of getting behind that wheel. They’ll recognize the responsibility it entails.

And they’ll understand the need for a multidimensional approach. They study their manuals to memorize the rules and best practices of driving, and they know they’ll need to drive an actual car under supervision to finish the process. They understand the theory-practice dynamic at play. So they’ll recognize that here too.

To earn their AI driver’s license, students should know how to…

1. Choose a Destination: No one learns to drive so they can sit in a parked car. Reflect, then pick a meaningful goal (personal, academic, or professional) to head towards.

2. Learn to Drive: Master the AI tools and skills you’ll need in college and careers. Then use AI to help you achieve the goal you set. As you do, learn how to adapt to each new model and keep up with a rapidly advancing field.

3. Open the Hood: Pause at key moments to examine how Gen AI works, what its current limitations are, and where various trendlines predict the tech is going next.

4. Reflect on the Rules of the Road: Collectively ask the big questions AI raises and start to propose answers. Know when to turn AI off.

This approach to AI literacy helps students shift from being passengers in the AI era, carried along by the choices of others, to being drivers of their AI experience, able to evaluate others’ choices and then make their own.

It positions AI as a car students can get in, if they choose, and use to help them get where they want to go in life. Once they get there, they’ll also know how to park the car, get out, and get on with living.

Upcoming posts will get into the details of each framework element and what those look like in my classroom and others. For now, here’s a vignette to distill the whole framework into one 45-minute period:

Students start by reflecting on their emerging sense of life purpose, first in notebooks and then in small-group conversations. They then open a custom AI bot programmed to have a 10-minute conversation about which careers might align with their purpose and how AI might soon be used in those careers. Halfway through, the teacher pauses the class to show an example of how one student is iterating their prompt phrasing to get better answers. Students then try that themselves. At the end of the conversation, they all ask their AI, “How did you come up with your ideas? What role did your training data and probability play?” They then close the AI and reflect in small groups about both careers that might interest them and their newfound thoughts about AI. The class ends with quiet, handwritten answers to these questions:

1. What career interests you most right now? How does it embody your purpose?

2. When might AI be used effectively in that career? When will you be more important?

AI literacy should give students agency as much as it gives them access. It should teach them more about the questions AI poses than about the ones it can answer. So, maybe counterintuitively, AI literacy should depend more on whole-class discussions than individual AI conversations.

---

My first long drive in that Camry was from Maine to Manhattan.

I remember how much fun it was to feel like I could drive on my own, guided both by my judgement and by the lessons my parents and a very patient NYC driving teacher had taught me.

I also remember how scary one particular merge onto I-95 was somewhere in northern Massachusetts. A truck was coming up fast and I didn’t realize at first how much I needed to accelerate. I realized quickly!

I sped up to finish the merge and then shifted down into a speed that felt right for me and was within the limit set by our society. I settled into the flow of traffic, joining all the other drivers around me, each of us using our machines for individual purposes but aware we shared a common roadway and were responsible for each other’s safety. Each of us making individual decisions guided by collective laws and norms. All of us heading somewhere we chose, knowing when we got there we would turn off the engines because our focus was on where we were going, not the machines we were using to get there.

In other words, I became a driver.

It’s time for our students to do the same.