I Have an AI Addiction

I was the kid who never played video games. My friends would spend entire weekends on consoles and I'd watch from the couch, bored and a little judgmental. The idea of sitting in one place for hours, staring at a screen, doing something that produced nothing tangible in the real world, it seemed like a waste of a life. I tried a few times, was terrible at most of them, and decided games weren't for me. That became part of my identity for decades. I'm not a gamer. I don't get the appeal.

Now I'm 40 years old, running three AI coding agents simultaneously across two monitors, burning through roughly a billion tokens a week across Lovable, Replit, Cursor, Claude Code, and half a dozen others, and I haven't moved from my chair in four hours. The irony is not lost on me.

There's a concept in game design called flow state calibration. The best games keep you in a narrow band between two failure modes: too hard and you quit in frustration, too easy and you quit from boredom. The sweet spot is right in between, where the challenge is just barely beyond your current ability, close enough that you believe you can reach it, far enough that you have to stretch. Game designers spend years tuning this. Difficulty curves, adaptive enemies, progressive unlocks. The whole architecture exists to keep you in that narrow band for as long as possible. Vibe coding does this naturally, and I think that's why it has its hooks in me in a way nothing else has.

For fifteen years as a CEO, I context-switched every thirty minutes. Meeting to meeting, decision to decision, fire to fire. Somewhere in those years my brain rewired itself around interruption. I stopped being able to sit through movies. Books took me weeks when they used to take days. I have ADHD, which didn't help, but the constant switching trained my attention span down to something that could barely hold a single thought for the length of a conversation. My partner and I would start a film and twenty minutes in I'd be on my phone, not because the movie was bad but because my mind physically couldn't stay in one place.

Vibe coding broke that pattern. Three hours pass and I don't check my phone, don't open a browser tab, don't drift. Three hours. For someone who hasn't sustained attention on anything for that long in fifteen years, that fact alone is worth sitting with. The reason is the same reason gamers can play for eight hours straight: the challenge level never stops shifting beneath you.

One moment I'm describing a feature in plain language and watching it materialize on screen, which feels almost effortless, like the early levels of a game where everything works and the world is generous. Then I hit a wall. An agent rewrites a component I didn't ask it to touch. The login flow breaks in a way I can't diagnose. Two agents are editing the same file and neither one knows the other exists. Suddenly I'm in a boss fight, except the boss is a hallucinating model that keeps telling me the issue has been resolved while the screen is full of errors. Easy, hard, easy, hard, impossible, breakthrough, easy again. My brain never gets the chance to disengage.

Then there are the reward loops. In games, the dopamine hits are engineered: the coin sound, the level-up animation, the treasure chest opening. In vibe coding, the rewards are real. Earlier this year I built a photo sharing page for our wedding. My partner has a background in design, spent years at Canva, and in our household she's always been the one with the eye for visual things. When I showed her what I'd made, she went quiet for a second, then started scrolling through it slowly. You made this? Our families used it, hundreds of photos went up, thousands of people viewed the site. That hit different than any level-up animation. That was a reward I could feel in my chest. Every feature that works, every bug that finally surrenders, every moment where I look at my screen and think that thing exists because I described it, the feeling accumulates into something that keeps pulling me back.

The losses are real too, which is part of why it works. When a context window expires mid-conversation and I lose four hours of shared context with an agent, the frustration is sharp and immediate, like dying on the final level and having to restart from a checkpoint three stages back. When I deploy a build and something breaks that was working ten minutes ago, the feeling is indistinguishable from losing a match I was sure I'd won. The emotional stakes aren't manufactured. They're mine.

A few weeks ago my partner and I were making dinner, and I told her I was in a good mood. She didn't look up from what she was chopping. It's not because of me, she said. It's not because of the weather. It's because you're addicted to vibe coding. My first instinct was to push back. Addicted felt too strong. Addictions are destructive. This was productive, creative, useful. She just looked at me. She was right, of course. She usually is about the things I'm slowest to admit.

And there's a multiplayer mode. Every Friday, my AI studio team does a show-and-tell. Each person shares one thing: something they built with AI that week, something they learned, or something they struggled with. Last week someone showed a tool they'd built for tracking surf lessons. Beautiful front end, completely broken when anyone other than them tried to use it. The whole room laughed, not at them, with them, because every person in that room had shipped something broken and felt that specific combination of pride and mortification. The show-and-tell is the social layer that makes the whole thing stickier. It's not a solo game anymore.

Here's what I keep coming back to.

There are roughly 3.4 billion gamers on the planet right now. Nearly half of humanity. That number took four decades to build, from Pong in 1972 to the first billion around 2010, then accelerating to where we are today. The gaming industry figured out something profound about human psychology: give people a finely tuned loop of challenge, reward, and progress, and they will show up for billions of hours every single week. The entire industry, now larger than film and music combined, is built on that insight.

The thing nobody is talking about is that vibe coding has the same loop. The same flow state calibration, the same dopamine architecture, the same progressive difficulty, the same social dimension. The difference is that gaming attached those mechanics to virtual outcomes. Points, levels, buildings that exist only inside the game, skills that reset when you turn off the console. Vibe coding attaches them to real outcomes. Real tools. Real products. Real solutions that ship to real people.

Gaming took 40 years to reach 3.4 billion people, and all it needed was a screen and a controller. Building with AI needs even less: a browser and a problem worth solving. The most addictive game loop ever designed is now attached to creating real things in the world, and the barrier to entry is an internet connection. The nurse who builds a patient tracking system. The teacher who builds an attendance tool. The restaurant owner who automates her scheduling. Each of them experiencing the same pull I feel at 4am, the same one more run compulsion, except the thing they're building actually ships.

I spent fifteen years watching talented engineers at my company stay late not because anyone asked them to, but because the problem had its hooks in them. That feeling used to be locked behind years of technical training. The wall is down. And what's on the other side isn't just the ability to build. It's the addictive pull of building itself.

It took gaming four decades to capture half the planet with loops attached to nothing real. I think building with AI reaches a billion people in a fraction of that time, because the loops are the same and the output actually matters. What happens when a billion people feel this pull?

Every other addiction I can think of leaves you with less than you started with. Less money, less health, less time. This one keeps leaving me with more. More tools that work. More problems solved. More proof that the wall I spent fifteen years standing behind was never as solid as it looked.