FIELD NOTES · LIVING WITH AI
MADE WITH AI, IN THE OPEN
This essay practices what it argues. It was dictated and edited by a human and built with AI, nothing hidden. The ✦ spark marks each AI technique that went into it, and we used all of them. Hover or tap any mark.
✦ = an AI technique used to make this essay · hover or tap a mark to read it
This essay takes its title from the question everyone in education seems to be asking right now: is using AI to do schoolwork cheating?
There are funny videos online of students sitting down with their laptops, reading an essay prompt from their teacher, prompting AI to write the essay, submitting it, closing the laptop, and being on their way. The student is a middle man of prompting between teacher and AI. It is an absurdity that undermines the whole activity.
The teacher's prompt is trying to get the student to do the act of writing an essay, to produce evidence of their understanding of a topic and their ability to communicate it. Even though the output might be correct, and increasingly AI's outputs are pretty dang good, having AI produce the essay is not evidence of the student's understanding, or of their ability to communicate it. If AI went away tomorrow, that student might not be able to produce a similar essay. And that seems to be a problem: what we really want is for the student to be able to produce that essay without AI doing it for them.
And notice: there is already an interesting involvement of AI in education, well beyond cheating. The main participants, the state, the teacher, the student, the employer, are all using AI to understand what is happening in schools and to take part in them. One way or another, AI has already become a co-creator of the educational experience.
But even before AI made this prompting chain a possible, cheating was a problem. If a student paid another student to write the essay, or went online and paid somebody to write it, we would call that cheating. So AI is not uniquely causing the problem. But it is exacerbating it, because it has made it very, very cheap and easy to not write the essay, or do the other performative paperwork of schooling.
But specifically, why do we care whether a student actually writes the essay or just prompts AI to write it? Some reasons:
One reason is that schools are like weird extensions of HR departments. Over the first 18 to 22-plus years of a student's life, we are trying to evaluate what they know and what they can do, so we can sort them into roles and professions and help employers find the right people to do the job. We really want to know who knows what and who can do what. It seems to matter a lot. If students are cheating en masse, the evidence of who is right for a job is compromised, and our ability to recognize who should do what, and why, erodes.
Another reason is that the activities of school can shape you. Having to show up every day, five days a week, and be there on time, ready to absorb knowledge, to compose that knowledge into the ability to write essays and do equations and solve problems, are all pretty good things to learn how to do. And you learn to do hard work. You learn self-discipline. You learn to set a goal and reach it through effort and grit. School is an environment where students learn to do this. And if you use AI to complete your assignments, that undermines this formative work.
The analogy I like to use is that this is similar to going to the gym and having a robot lift the weights for you. That would be incredibly self-defeating. You would not leave having achieved your goal: working your body, getting stronger, building cardiovascular capacity. Similarly, if you go to school and have AI do everything for you, you have not achieved the formative good of school.
So we have two things AI short-circuits: the meritocratic sorting goal, and the formative goal of school.
Given that students are using AI to try to "cheat" their way through school, or at least their use of AI is making it difficult to really understand how effective schooling is at imparting know-how and identifying who should do what and why, it is worth asking why they are doing it. Worth asking the students themselves, of course, but this essay may give us a framework for exploring it with them.
This is hard to answer simply. Which students are we talking about? There is a vast diversity of experiences students have. There are over 13,000 school districts in the U.S., each with its own school board. There is a huge disparity between the poorest-funded schools and the wealthiest: the resources students can access, where the school sits relative to home, whether they walk or get dropped off or ride a 40-minute bus from the rural countryside, what is actually taught (which, despite decades of standardization, still varies), who they go to school with, the home lives and economic resources of their neighborhoods. To speak about school in the U.S. is not to speak about a monolithic thing. This won't be an exhaustive diagnosis. But it is still the right question to ask.
Here is a hypothesis. When the ability to cheat becomes extremely easy and cheap, students cheat because we have made it so totally unobvious why doing the things asked of them is actually good for them. We adults may think that going to school is like going to the gym, and it should be obvious to students that they shouldn't use AI to cheat because they are in fact cheating themselves. But the fact that students are using AI to cheat to me suggests that it is not obvious to them how doing school is to their formative benefit.
Why might it not be obvious to students that doing school is formatively good for them? At least one unifying experience all students face is an extreme, competitive, meritocratic pressure. The one thing No Child Left Behind and Common Core and the rest have reliably produced is a Chinese gaokao-like pressure to succeed along bureaucratic performances of intellectual feats. If the most important thing about school is winning that performance, and you suddenly have a tool that produces the very evidence adults are looking for, it makes a lot of sense to use it.
What the adults say they care about is that you are actually the one who could write the essay. But the way they determine that is by grading the artifact, so it is the production of a correct artifact that matters. And AI is very good at producing correct artifacts.
In these circumstances, we have not made it obvious to students that:
- writing that essay is in fact good for them, because learning to shape a thought into words is part of becoming someone who can move other people
- doing that homework by hand is in fact good for them, because that is where the capability actually takes root, and with it the agency to become who they want to be
- studying hard and then performing closed-book, from memory and skill, is in fact good for them, because learning to embody knowledge and skills gives shape to who you are, even if later you augment that will tools
We have made it difficult for ourselves to organize schooling around the formative good of students. The meritocratic competitive eclipses the formative goods, and it is a symptom of a deeper set of goals. We have asked schools to deliver justice in our society, not just education. The way we made schools responsible for social justice is by using them to distribute "opportunity." If every student receives an equal opportunity to potentially succeed, then our economic and political system, which is always answering who does what and who gets what and why, is justified. Even if the outcomes are very unequal, who can complain, since they were given the same shot as everyone else? So we placed the work of justice on the student, the teacher, and the family to take advantage of a competition and try to win, because we've abandoned doing justice within the realm of politics. There is little recourse outside of winning this competition for hoping for achieving a better and more comfortable life in our society. This is why the stakes in education feel so high and the competition is so fierce.
This greatly distracts from doing right by the formative potential of students. To the extent that we care about the formative good of students, we are making a bet: that the formative good of students is aligned with, and served by, the meritocratic competition we have set up. Fortunately there is some overlap. Spending 18-plus years competing in reading, writing, math, and science does have some good formative effects. You can get disciplined. You do learn a lot about the world. But it is a narrow curriculum toward a narrow set of competitive goals:
- do really well in school
- get high grades
- outcompete your peers
- get recognized by colleges
- pick difficult majors aligned with lucrative careers
- outcompete your peers
- land the jobs
A snag. There are not an infinite number of high-paying jobs, and we now compete for them with students around the entire globe. At different stages, some students begin to feel they are not going to win. If the competition is justified in terms of winning it, and you see there is not enough space at the top for you, then all that work starts to feel a lot less justified. This is where we fall back on "but it's good for you," argument - do it so you won't become lazy, so you become disicplined, using the formative goods of school as a secondary, conciliatory justification.
The competitive version sends everyone racing up one narrow chute toward a single prize at the top. And the climb is bought: private tuition, daily tutors, summer camps, test prep, admissions consultants, each rung another bill. A few reach the landing. Most fall back, some into lesser buckets, many onto the ground.
We talk of AI as cheating, but look at what we have come to treat as ordinary, because of high stakes schooling. We use performance-enhancing drugs in school, ADHD medications for example, to win. We use money to buy an edge, paying for tutoring services like Kumon. We will chemically and financially optimize a child to take the prize.
Students recognize that who they want to become may be inaccessible to them, not because they couldn't do it, but because they might not be able to win the competition to get access to the resources and support to do it. They might also see the alternatives offered to them are not rewarded by society. We have put the inequalities of this competition more and more in each other's faces. The rewards for winning grow graver and more public, raising the stakes, and raising the psychological blow of realizing you will not win. So aligning your formative efforts toward the alternatives can feel like getting second place from the start, or third, or last. And so that formative work just doesn't feel as good either.
This is a difficult context in which to center the formative work of schooling.
I want to and underscore what an enormous disservice this context does to the formative potential of students. If we could reverse the emphasis of schooling, and first and foremost take care of the formative good of students, and only secondarily look for evidence (potentially competitive evidence) of who should do what, schools might serve students far better. And we may even succeed in addressing the AI cheating problem. We have discussed the WHY behind students using AI to "cheat" at school, and the struggle to justify not using AI in a hyper meritocratic context. But, now I'd like to examine whether using AI in and of itself as a part of education is a bad thing.
To do that, we need a more accurate picture of what education even is, and where a technology like AI fits into it.
MOVEMENT ONE · WHAT EDUCATION ACTUALLY IS
“You can’t do much carpentry with your bare hands, and you can’t do much thinking with your bare mind.” BO DAHLBOM AND LARS-ERIK JANLERT, QUOTED BY DANIEL DENNETT
The point of that line is that human intelligence is always augmented by our use of tools and our relationship to the environment. Too often we think of intelligence as an innate trait in a person, their identity: they’re dumb, they’re smart, they’re clever, they’re capable. This is the same picture that lets us treat school as a distributor of opportunity and a producer of evidence of who deserves to do what.
There’s some truth that skill and capability become embodied, and grow and diminish over a lifetime within a person. But it is more so that person’s ability to organize their life and their world around them, their resources, food, clothing, tools, relationships, location, that unlocks those capabilities and enhances them.
A biologist named Jakob von Uexkull gave this a name a century ago: the Umwelt. We don't experience the world as it objectively is, in total. We experience a slice of we can actually perceive and act upon. Perception and action close into a single loop, and that loop is a person’s world. A tool does not simply allow you to do something. It widens the loop you can reach, and thus it widens the world you have access to. And that has all sorts of consequences for who you can become.
Picture a person alone on a deserted island. No tools, no others, no resources. Strip all of it away and a startling thing shows up: there is almost nothing they can do. Who you are and what you can do are very nearly the same thing, and both of them are on loan from the world you are integrated into.
A person is a heteronomous thing. Our will and our agency are co-determined by the world. So to govern ourselves and grow in capability, we have to govern our circumstances.
Education is every human’s potential and need to grow and change, and to direct that growth in concert with the environment, with some intention of becoming somebody. I need to learn to farm so I can grow food and eat well: what I am really doing is organizing myself into someone who can farm. A farmer. Self-formation, in concert with the world. That is the educational landscape, more or less from first principles.
MOVEMENT TWO · TOOLS, AND THE TURN
The lever of self-formation has always been our tools. The minds of humans 300,000 years ago to today have not substantially changed, and yet the things an individual can accomplish today are incredible, and the things people can accomplish together are incredible. There were no spaceships even a couple hundred years ago. Why not? People were potentially just as smart.
What is happening is an increasing sedimentation of tools and ideas, stored in things like books and other media, that accelerates how quickly somebody can become highly capable, and organize with others to achieve more and more complex feats.
Tools are the layer between our will and our ability to organize the world. They extend our ability to organize our environment, which then organizes us, and so on, in repeat. So when a new tool comes along, like AI, the way to evaluate it is its formative potential.
So AI is not a uniquely corrosive thing. It is the newest, most powerful entry in a long line of tools that widen what a person can perceive, do, and become. The only question worth asking of it is the formative one.
Our educational capacity, as humans, is the ability to make a hypothesis, test it, look at the evidence, evaluate the hypothesis, and repeat. We are always doing that, no matter what. It is just our way of being. Right now students aim that engine at the games of school. It is their formative power, the tools are already in their hands, and it could be aimed at the world. AI lets you run that loop on almost anything, at speed.
Modern education wrapped a school around that human capacity. We decided to take care of each other and each other’s children, to create organized time, space, and resources, so kids could get oriented and begin organizing themselves into the people they needed to be: citizens, with whatever skills the economy of the day asked for. We tried to shape ourselves through a public effort.
You could think of the formative potential of a student as an underlying capacity, capable of a vast range of learning, work, growth, and change. To use the technologly of our moment as analogy, a school is a wrapper we put around that capacity to direct it toward certain outcomes. Right now it is extremely emphasized on a narrow set of outcomes, along a narrow set of competitive evaluations. But we don’t have to organize it that way. If we changed how we organized school, we would help students relate to that formative capacity in new and potentially more obviously compelling ways. We can shape that wrapper in an infinite number of ways.
So what are the formative goods? Hard to say in full. But broadly: a healthy body, learning to take care of yourself, to eat and move and sleep and work well, a level of self-control held intentionally. Similarly the mind: practices that help you become effective at achieving your goals. And the heart: learning to be secure, to find love in its many forms, to be in community, to participate in efforts meaningful to you and others. If we help students organize themselves, in concert with the world, around these pillars, then we have succeeded.
If we secured these formative outcomes for all students, the competition we also need would probably work even better. The stakes would be less grave, so there would be less incentive to subvert it, by cheating, by social capital, by money buying access to a winning slot. People could ask themselves more honestly, “Is that something I actually want to do?” because the binary of succeed-or-fail would be greatly diminished.
If I want to be great at communicating my ideas, if I want to melt hearts with a poem, if I want to find value in unusual places and build a business, if I want to make plants more heat-resistant as the planet warms, then who do I need to be, and how do I become that person? You set out on a path of discovery of yourself, the world, and the problem. And through that journey, you grow and shape yourself into the person you want to be called.
So, to not do a rug pull: the answer to “Is using AI in school cheating?” depends a lot on how you have organized school. If using AI erodes a formative practice, then yes. If using AI erodes a merely performative practice, then who gives a shit, except in the rare, high-stakes cases where we genuinely need good evidence that someone can do a thing.
So instead of asking “Is using AI in school cheating?” we should be asking: is our current wrapper around our educational capacity of students, their formative capacity, suited to the technologies we now have? And, the sharper version:
THE FORMATIVE TURN
The moment in human history when the goods of self-formation become the center of our culture, politics, technology, schooling, and economics, as the justifying bottom line. We start asking, of any new technology or policy or medium or culture or art or product, anything: who does this make us?
This essay, made the way it argues for: dictated and edited by a human, drawn with AI, evaluated for clarity and iterated with care. An exploration of how AI can help us better communicate with each other to see reality more clearly.