I was wrong about the internet.
Not about the technology. The technology did everything we thought it would. I was wrong about what it would do to people. Not eventually. Fast.
Within a decade, recommendation engines were optimizing for outrage because outrage keeps people scrolling. Teenagers were being served content that told them they weren't enough, not because anyone decided that was okay, but because the engagement numbers looked great. Designers who believed in what they were building were shipping addiction mechanics. Platforms built for connection turned out to be better at isolation.
I believed. I was completely, confidently wrong.
I'm telling you that because it matters. Being wrong about the internet isn't a disqualifier. It's the tuition. I paid it. I know what it feels like to watch certainty dissolve. That's not weakness. It's the equipment.
I also know what it feels like when the field you're standing in changes underneath you and nobody names it.
So when I tell you I think AI is different, you know what that's worth.
What I'm making is a smaller, more specific claim. And I want to show you where it came from.
01 The kid who didn't fit the room
It's 1995. Seventh grade. The class assignment is to present a new product. My classmates present skateboards and snowboards. I present a laptop I designed.
Not one I owned. One I built in my head from specs I'd been reading in Computer Shopper and PC Magazine every month. 486 processor. 32 megabytes of RAM. Notebook form factor. I knew exactly why none of it was practical. The 486 was too big, too hot for a notebook. I designed past the constraints anyway.
Nobody got it.
My teacher, Ms. Farrell, got it enough. I don't think she understood what I was presenting. How could she, in 1995, with a 13-year-old describing a laptop that literally couldn't exist yet? But she made space for me. She made sure the room made space.
I felt seen. Not because anyone understood what I was saying. Because someone decided the room had room for it.
I've never forgotten her.
Years later I'd encounter a Star Trek principle and recognize something I'd already been practicing. IDIC. Infinite Diversity in Infinite Combinations. The person whose reference pool points somewhere else. The kid who can't connect with the room. The designer who keeps asking: what about the person the pattern missed?
I didn't adopt that philosophy. I recognized it. The label arrived thirty years after the instinct.
That's how absorbed knowledge works: it forms before it has words, the way a first language does. Not through instruction but through immersion. Trek wasn't curriculum. It was atmosphere. That's why the recognition felt like remembering, not learning.
Same with Ex Astris Scientia. From the stars, knowledge. Go toward what you don't understand. Come back with it. I did that at 13 with processor specs I had no reason to know. I did it in 2008 when I went looking for an IT director nobody told me to find, learned a proprietary file format over a weekend, and eliminated a data entry process that had run for years.
The motto didn't give me a method. It named one I already had.
Star Trek wasn't popular for 60 years because it had good special effects. It kept insisting the future could be better than the present, and that getting there required treating every person as worth designing for. I absorbed that over 35 years before I connected any of it to my work.
The language arrived. The instinct was already running.
Which is why, when AI arrived, I wasn't looking for the next evolution of humanity. I was looking for the seam.
02 What design has been missing
Commander Data spent seven seasons of Star Trek trying to become more human. Studying emotion. Attempting humor. Writing poetry. He knew what he lacked. He kept reaching.
The machines we're building today look like they're doing the same thing. They're not.
Data was reaching toward the exceptional: the specific quality of being human he admired and couldn't quantify. Today's models are processing the average. Every model is a compression of its training data, and the training data is overwhelmingly drawn from the center of every distribution: the most common user, the most common workflow, the most common expression. The output isn't aspiration. It's a median. The bell curve written in the data.
Data knew the gap was there. The models don't know a gap exists.
That's not a consolation. It's a job description.
In 1977, an architect named Christopher Alexander published a book that asked it directly. What makes a designed thing alive? Not usable. Not beautiful. Not innovative. Alive, coherent, human, right, in a way some spaces have and others don't, and the difference is real and traceable if you're willing to look.
Architecture got that question. Typography got it. Product design, interaction design (the discipline most of us actually work in) never did. What it got was methodology. Double diamond. Design thinking. Journey maps. These are tools. Good tools. But they don't answer the question: what makes this good, and for whom, in a way that survives the next tool shift?
We've been practicing without a foundation. For a long time we could get away with it, because the distance between "good enough" and "good" was hard to see.
AI closes that distance. Fast.
A model can produce work that passes every usability heuristic, satisfies every accessibility requirement, and clears every stakeholder review. In minutes. What it cannot do is ask the right question before starting. It cannot feel the difference between a solution that fits the pattern and one that accounts for the person the pattern missed. It optimizes for the center. It has no philosophy of the edges.
The gap between work that passes and work that's alive is the one a practitioner needs a foundation to defend.
03 The reckoning
I ran the research on what happens historically when execution commoditizes in knowledge professions. Legal, marketing, design. The pattern was consistent and not comforting. Execution commoditization doesn't automatically produce a judgment premium. The value accrues to platforms. Practitioners who thought craft would protect them found out it didn't.
I sat with it. Then I pushed back.
The research was about automation. Automation dismantles the execution layer. AI isn't downstream of the judgment process. It's inside it. A reasoning partner operating at the level of the thinking itself.
I used AI to sharpen a conviction that AI's own research was arguing against. The act was the argument.
If the traceable call is the thing (the documented reasoning chain, the user story, the moment of insight captured in a journey map) then the bet fails. That's just training data. You don't protect yourself by making your judgment more legible. Legibility is the vulnerability, not the moat.
The thing that doesn't compress into documentation is the moment before documentation starts. The question that doesn't have a name yet. The IT director nobody told me to find. The thing someone says off-script in a session that changes the direction entirely. You catch it because you were in the room. That's pre-legible. It doesn't exist yet when it matters. You can't train on something that hasn't happened.
The internet shaped what you saw, what you bought, what you believed. It was consequential. It was also, at the end of the day, a surface. You could close the tab.
AI is underneath the surface. It's in the hiring algorithm that never shows you the job posting. The loan decision that doesn't explain itself. The medical diagnosis. The bail determination. The content moderation call that decides whether your voice exists in public. Being built right now, by people who may or may not have a foundation that tells them what they're building toward.
I'm not pie-eyed about this. I watched what the internet did. But if the tools get better at the center, the edges don't disappear. They become the work. The work is never finished, because the moment you name an edge it becomes the center, and a new edge opens. You don't get harder to replace. You stay in the current. The day you stop going toward the thing that isn't named yet is the day you're standing on ground the tools already reached.
That bet might not pay off. My father-in-law says everything is cyclical. He's probably right. If the cycle closes badly, I want to have been wrong on the right side of it. On the side of the humans. On the side of the person the dropdown didn't have room for.
That's the right side. Even if I'm wrong.
04 The founding
My son Ezra is four. He's adopted. He doesn't play with toy buses. He knows the routes.
The 66 Chicago, the 65 Grand, the 29 State. They all terminate at Navy Pier. He knows the streets they traverse. He knows their far terminals at the other end. When we board a bus, he asks for the final stop, because he knows the final stop tells you the direction. When the 65 Grand runs to Nordica instead of Chicago, he clocks it.
At home, with his toy buses: "66 Chicago to Navy Pier, this is Franklin, transfer to Brown line."
He's four. He has no reason to know any of this. The same intensity I had at 13 with a processor spec I had no reason to know.
I didn't pass that to him. He arrived with it. Not the buses. The obsession. Every kid does. The instinct isn't inherited. It's a species-level constant that gets either held open or slowly closed down by the world around it.
As a parent, as a designer, as a leader: your job is not to install the instinct. It's already there. Your job is to be Ms. Farrell. Make space for the thing you don't fully understand yet. Hold it until it finds its language.
I spent 35 years building toward a philosophy before I had a name for it. This wasn't the first time I tried to find it. Blogs, journals, YouTube, whatever container seemed right at the time. I kept running out of road before I got there. Not failure. Reconnaissance. I was practicing without knowing I was practicing.
· The name is Tanagra
There's a Star Trek episode where Picard encounters a captain whose language nobody can translate. Not the crew, not the universal translator, nothing. Every attempt fails. So the alien captain does the only thing left: he beams them both to a planet, alone, with a creature neither of them can fight by themselves. In surviving it together, they finally understand each other.
His language wasn't words. It was shared experience, compressed into reference. "Darmok and Jalad at Tanagra" doesn't describe the meaning. It is the meaning. Every event, every emotion, every instruction, encoded in an utterance that only works if you've lived through the events that generated it.
That's the model. The community generates the events. The events become the stories. The stories become the language. And the language isn't a wall you hide behind. It's a current you stay in. Alive only while the community keeps moving.
This is the first event. You're already in it.