An Open Letter to the Class of 2026

The TL/DR: The video summary is just a few minutes shorter than actually reading the article, but Google’s NotebookLM does a decent job it even if it always produces predictable, bland, “canned soup” output.

You’ve probably been told a lot of things about AI this year — by your teachers, your parents, the people in tech who built it, and people in institutions who are afraid of it: everyone has an opinion, a policy, a warning, or a sales pitch. You’re told not to use it because it’s cheating; you’re told to use it because it’s just a tool; you’ve watched your school issue policies and sat through assemblies about academic integrity that left the real question completely untouched. The real question isn’t whether you use AI — it’s what happens to your thinking when you do (spoiler: it’s not all bad).

I’m not a teacher, or school administrator, or a tech guru; I’m a grandmother — and also a researcher and retired executive who has spent decades studying how humans make meaning and develop potential: how we learn, how we grow, how we become the people we’re capable of being. That combination gives me a view I don’t hear much in the current noise about AI and education. I think it matters — not because it will help you pass the next exam, but because it might help you become a better, more confident human: the kind of person who knows what they think, knows what they’re for, and can tell the difference between those two things and borrowed versions of both.

So before I tell you about four students I know, I want to tell you something about how AI language actually works — and how it’s different from what happens when a human being puts language into the world.

I make soup from scratch, and my grandchildren can tell the difference between my soup and soup from a can — not because I use better ingredients necessarily, and canned soup is fine. Canned soup is convenient; it does the job. But it was engineered to be acceptable to as many people as possible, by a process that has never met any of them. My soup is different, not because it’s fancier, but because I made it: my hands are in it, my specific way of doing things, the way I learned from my mother and grandmother, the adjustments I’ve made over decades, the fact that I’m making it for people I know and love. You can taste that it came from somewhere real.

That’s the difference I want you to understand about AI and language.

AI learned to produce language by processing more text than any human could read in a thousand lifetimes — what it learned are patterns: what words tend to follow what other words, what structures feel coherent, what most people seem to mean when they write about most things. It learned the outside of language remarkably well, and it produces canned soup: fluent, coherent, optimized, and made for everyone, which means made for no one in particular.

Human language, when it’s actually working, is something else entirely; it carries the person who made it — not just their ideas but their encounter with something real: their body, their history, their specific way of moving through the world. That’s why a piece of writing can make you feel something before you understand it, and why you can tell when someone means what they’re saying and when they don’t, even when the words are identical. The difference isn’t in the words; it’s in what’s behind them. Human language is tears and laughter, it’s art, and the smell of soup someone made because they love you. AI has never loved anyone; it has never cried, it has processed everything and encountered nothing.

This is why the real question about AI and your education isn’t whether you used it — it’s whether who you actually are shows up in how you used it: your particular history, your specific way of seeing, the person who has lived your life. (Side note: schools and teachers don’t know yet how to evaluate for that, and so-called AI detectors don’t work.)

A word about what kind of AI I’m describing, because the landscape is moving, and this letter should be honest about that. The AI available to you right now learns from text and generates text (and bad graphics, sometimes)— but that’s already changing: AI is becoming more capable of acting in the world, not just describing it. It can make decisions and pursue objectives in ways that go well beyond responding to a prompt, so labeling AI as “just a tool” is incomplete at best. Serious conversations are happening about whether AI might develop something that functions like emotional response. What AI actually is, not just now but as it evolves, is a question worth its own serious examination, one I won’t resolve here. What I can tell you is that the question I’m asking in this letter isn’t really about any particular version of AI; it’s about you. Whatever AI becomes, the question of whether you showed up to the work as yourself — whether who you actually are is still recognizable in what came out, or whether AI quietly replaced you with something more average and acceptable — doesn’t change. As AI gets more sophisticated, you still have to decide what you’re bringing to it.

Now let me tell you about four students I know. One of them is probably closer to you than you expect.

Steven creates graphic novels — a complete series with a hero, a villain, and a world built from scratch with its own internal logic and mythology. He posts about it online with a “spoiler-free” caption, because he’s already thinking about what his reader deserves. He built this world at fifteen.

He’s also funny in a specific way — not trying to get the whole room to laugh, but sharper, operating at a frequency only certain people can appreciate. At thirteen, it was fart jokes. As he has grown, his unique way of seeing things has grown, too. He finds the angle nobody else found, says the thing the room wanted left unsaid, refuses to perform a politeness he doesn’t actually feel. How he brings language to the world is more precise and more himself with every year.

I want to tell you how Steven actually uses AI, because it’s the clearest example I know of the difference between canned soup and cooking from scratch.

He uses AI to refine pacing — not to generate story but to stress-test what he’s already built, asking it to find the gaps in his narrative logic before a reader does. He uses AI to challenge his own work; he maintains full ownership of every creative decision. He could have AI generate his artwork, write his scripts, have a publishable product tomorrow if he handed the cooking over — and he won’t, because he knows the difference between his work and work that merely looks like his. He’s skeptical of AI in ways that go beyond preference to something more like principle: the work has to be his, or it isn’t worth making.

This is what intentional AI use actually looks like, and it’s worth naming clearly because the noise around AI in education tends toward two useless poles — blanket prohibition and uncritical adoption. Steven is doing something more sophisticated than either: bringing AI into the specific places where it serves his thinking, and refusing it everywhere it would substitute for that thinking. The filter is always his. The work is always his—the soup is always his.

If that’s you — if you’re already navigating this with that kind of intentionality — trust it and keep developing it, because the AI you’re using now is only the beginning of what you’ll encounter in your working life. The habit of knowing what you’re bringing to it matters more with each version that arrives.

The first time Mark visited a theme park as a child, the highlight wasn’t the rides — it was figuring out how the soap and towel dispensers worked. That is not an unusual story for Mark; it’s a completely representative one. When something catches him, it’s the mechanism, the system, the chain of causes that makes a thing do what it does — Rube Goldberg machines, marble runs, dominoes, chain reactions of any kind. These aren’t hobbies so much as a natural language, the register in which his mind moves most fluently. He’s been learning to code and recently graduated from block coding to C, which is less a milestone than a natural next step for a mind that has always been drawn to how systems work from the inside.

Regular classrooms haven’t been the right room for him — not because he can’t think, but because a room full of people, unpredictable noise, and constant performance demands costs him so much to survive that there’s nothing left for actual learning. He’s not refusing to engage; he’s already full before the lesson starts. For Mark, capacity isn’t an obstacle, but the environment can be.

AI might be one of the learning spaces that doesn’t cost him everything to enter: it’s one-on-one, patient, it explains the same thing seventeen ways without making him feel strange for asking, and there are no other people in it. What he figures out there, he figured out — AI won’t be an obstacle to his thinking; and it won’t do the thinking for him.

If that’s you, genuine engagement feels like something: your mind actually moves, you go somewhere you weren’t before, the mechanism reveals itself, and something clicks. If that’s happening, AI is serving you; if you’re skating the surface and producing answers without anything real happening underneath, you’ll feel that too — a hollowness, like canned soup when you were expecting something more. Trust that signal; it knows the difference.

James is entering a computer science program this fall, and his mind moves through systems in ways that diverge from how most people think — finding patterns others miss, making connections others don’t see, reaching solutions the mean wouldn’t reach. This isn’t performance; it’s how he’s built.

His relationship to AI will be different from everyone else in this letter: for the others, the question is whether AI serves their thinking or substitutes for it, but for James, AI will be a major part of a career in computer science — he’s going to live and work inside it professionally, which means the risk is more specific and more consequential. He already knows this.

AI learned from an enormous amount of human-generated content, and what it learned is a statistical average — the way most programmers, across millions of examples, tend to encounter and approach most problems: the mean. James is not the mean. If he starts with AI’s solution, he’s inside someone else’s framework before he’s even found his own — opening a can instead of cooking. For most students, that’s a mild concern, but for James, it could mean four years spent becoming a very good modifier of average thinking, when he had the potential to discover and innovate something the average computer scientist would never have imagined or accomplished.

If that’s you — go in first, every time; have your own encounter with the problem before you look at what AI does with it, then bring AI in as a sparring partner and see where it diverges from you, and ask why. Sometimes it will be right; sometimes you will be right; and learning to know the difference, from the inside, is the most important thing four years of CS can produce. Your way of seeing is not an edge case to be corrected toward the mean — it’s precisely what the field needs most.

Allison doesn’t struggle with the system; the system works for her — essays, analysis, structured arguments delivered through exactly the forms academic institutions reward. She’s deciding between two or three competitive colleges, and by every conventional measure, she’s succeeding.

She’s also one of the most perceptive people in any room she walks into: she reads situations fast, calls things as she sees them, has a low tolerance for performance masquerading as substance, and trusts her own read over the social consensus — and her read is usually right. These aren’t quirks to manage; they’re precision instruments, a specific way of encountering the world developed through years of paying close, honest attention.

Here’s the question nobody has asked her: beneath the performance of mastery, what is she actually developing? For Steven, Mark, and James, their real intelligence shows up in what they make, how they think, and what they choose to know deeply. Sometimes, a traditional education institution sees it; sometimes, they have to work a bit harder to demonstrate their intelligence in ways the institution requires. For Allison, the institution sees her clearly and rewards what it sees; but what it sees may be the surface of something deeper that the performance of mastery has never been asked to reveal — possibly even to herself. No institution’s approval can substitute for self-knowledge.

AI can now produce exactly what she’s been producing — coherent, well-structured, institutionally excellent work, canned soup beautifully labeled. If she uses AI to generate more of what she’s already good at, she risks four years of college without ever finding out what she actually thinks when nobody is asking her to demonstrate what she knows.

That precision she has — the ability to cut through to what’s real, to be right even when it’s uncomfortable — is the product of her specific encounter with the world. It belongs somewhere specific: where clear-eyed judgment under pressure is the job, where being right matters more than being liked, where the work itself demands the kind of mind that doesn’t flinch from what it actually sees.

If that’s you, the question worth sitting with is: what am I actually like when I’m working at my best? Not performing — actually thinking. Succeeding at academic structure and format is not the same as knowing what you’re for, and knowing what you’re for is the most important thing the next four years can produce, if you let them. Don’t let AI answer that question for you; it can’t, because it’s never been where you’ve been, and it has never wanted anything.

Steven, Mark, James, Allison — four completely different minds, four completely different relationships to learning, to language, to AI.

But here’s what they share: each of them has something in their language that cannot be replicated — a specific encounter with the world, accumulated through a specific body, a specific history, a specific way of paying attention, that belongs to them and cannot be generated by any version of AI, however sophisticated. For some, the system could see it; for others, it couldn’t; but it was always there, underneath whatever the institution was or wasn’t able to measure.

The problem of work that looks like thinking without being thinking is as old as school itself — students have always found ways to hit the marks without their mind ever actually moving. AI didn’t create that problem; it gave it a very sophisticated can opener for canned soup. AI just made it faster, easier, and harder to see from the outside. Whatever AI becomes next, that problem will persist and probably deepen, which is why the habit you build now matters more than the specific version of AI you’re building it around.

The only instrument that detects the difference is the one you already have: the felt sense, from the inside, of whether you actually showed up — whether who you are is still recognizable in what came out, or whether AI quietly replaced you with something more average and acceptable. You’ve known that feeling your whole life — when something clicks, when you surprise yourself with what you think, when you know something is true before you can explain why, when the soup tastes homemade. That feeling is the most reliable thing you have; cultivate it, protect it, and bring it to every encounter with AI for the rest of your life. It’s the only thing a very sophisticated world cannot manufacture.

Class of 2026: the most important thing you’ll develop in the next four years isn’t a skill or a credential — it’s the ability to tell your own thinking from a borrowed version of it, from the inside, before anyone else can see the difference. That matters more, not less, as AI gets more sophisticated, and the world genuinely needs people who can do it. You already have the instrument — that felt sense of whether the soup is yours or came from a can. Trust it. Bring it. It’s the only thing a very sophisticated world cannot manufacture.

With love,

Dr. Grandma