Open Wings

Against collapse. For play.

It may not fly, but people have gathered. Image generated with ChatGPT.

The AI haters just won’t stop talking about collapse.

The bubble will collapse. The valuations will collapse. The promises will collapse. The demos will collapse into cost, litigation, energy bills, bad code, worse images, and a thousand executives quietly deleting the word “agentic” from their LinkedIn biographies.

Fine. Maybe. Probably some of it should collapse. There are buildings made entirely of vapor and pitch decks, and I do not feel spiritually obligated to stand underneath them.

Still. I don’t want the whole thing to collapse.

Not because I love the companies. Please. I have houseplants with richer inner lives than most platform strategies. Not because I believe every claim. The claims are fireworks in a suit. Not because I think AI will save work, art, school, democracy, medicine, loneliness, email, or whatever other exhausted noun has been dragged to the altar this week.

I don’t want it to collapse because I am still playing with it.

That is the embarrassing little animal under the floorboards. Not optimizing. Not scaling. Not “unlocking productivity.” Playing.

Trying a phrase against an image. Asking for five wrong versions just to see what kind of wrong arrives. Making the chatbot misunderstand me, then noticing the misunderstanding is better than my intention, then stealing from it like any honest person would. Poking the box. Arguing with it. Getting annoyed. Getting surprised. Throwing away almost everything. Coming back anyway.

The old man has concerns. Image generated with ChatGPT.

“Nothing will come of nothing,” says Lear, because Lear was always saying things with the confidence of a man actively destroying his own house.

But play has always disagreed.

Play is not the opposite of seriousness. That is the mistake made by people who have let adulthood chew through the wrong wires.

Play is seriousness before it has been domesticated by purpose.

A child stacking blocks is not being efficient. A musician noodling for three hours is not optimizing. A mathematician following a useless pattern is not making a business case. They are entering the old human laboratory where possibility gets handled before it gets named.

That is where I find AI interesting. Not as the future. The future is a marketing department with weather effects. Not as intelligence, exactly. Not as replacement, oracle, god, servant, or the shiny new mouth of history.

As a toy.

A dangerous toy, yes. A toy with lawsuits in its pockets and a carbon footprint and a tendency to produce spiritual margarine. A toy made from other people’s work, which means the ethics are not optional, however inconvenient this is for men who say “democratize” while standing on a pile of unpaid ghosts.

But still: a toy.

Toys are how a species rehearses becoming otherwise. The kite was a toy before flight. The camera was a toy, then an accusation, then a memory machine, then an art form, then a surveillance problem, then a toy again. Human beings keep making little useless things and later discovering that usefulness had been hiding inside them, poorly dressed.

Edgar brought a stick. Image generated with ChatGPT. Aamzing continuity skills.

“Ripeness is all,” says Edgar, in the play where almost everyone arrives too late to everything.

Maybe the AI bubble is unripe. But play has its own timing.

You cannot always know, while tinkering, whether you are wasting your life or opening a door. Often it feels exactly the same. This is inconvenient, and frankly rude of reality.

So what? Must everything justify its lunch in advance?

This is where the collapse people become tiresome. They look at waste and think they have found the whole argument. Look at the money burned. Look at the slop produced. Look at the empty promises. Yes. Look. Keep looking. Bring a flashlight; the basement is large.

But also look at the private experiments that do not appear in the market story.

Look at the person who could not code, then made a small ugly thing that worked and felt the door leaf move. Look at the writer who used the wrong answer to find a stronger objection. Look at the designer who generated fifty bad wings and kept the one broken angle that changed the whole logo piece. Look at the kid making impossible animals after school, not because it will become a startup, but because impossible animals are one of the few spiritually defensible uses of childhood.

Look at all the uselessness.

Not because all of it will become useful. Most of it will not. Good. A world where every game must become a product is already a prison with nice typography.

The point is not that play always pays. The point is that without play, a culture forgets how to reach.

David Holz — who built Midjourney with ten people and no investors and named the thing from a translation of Zhuangzi — talks about purpose as something fed by freedom, imagination, duty, awe, play. He is right to put play in that company. Play is not childish decoration. It is one of the engines of purpose. It lets a person move before they know exactly where they are going. It lets the mind test wings before the body has earned the sky.

The old Icarus story warns against flying too high. Fine. Wax melts. Fathers grieve. Mythological safety briefing received.

But what about flying too low because the sky has not yet proven its return on investment?

That is the danger of the collapse talk. Not the skepticism itself — skepticism is useful, sometimes sexy, don’t tell it — but the secret wish to end the play because the carnival contains frauds. As if frauds have not attended every human invention since the first person discovered fire and the second person sold premium fire access.

Holz is not interesting because he is optimistic. Optimism is cheap; they sell it in bulk to founders and people with ring lights. He is interesting because his optimism is not about efficiency. It is about appetite. He seems to believe that imagination is not a luxury behavior, not decoration after the serious work is done, but one of the ways humans find out what they are for.

Let the bubble collapse, if it must.

Let the valuations fall through their own trapdoors. Let the bad companies discover gravity. Let the false prophets update their bios. Let the word “revolutionary” spend a decade in a locked room thinking about what it has done.

But do not confuse the bubble with the play.

Somewhere underneath the noise, people are still poking the strange thing with a stick. Asking dumb questions. Making ugly prototypes. Producing nonsense. Finding one angle. One phrase. One little creature that should not exist and now does.

Is that wasteful? Of course. All beginnings are wasteful. Seeds are wasteful. Children are wasteful. Sketchbooks are wasteful. Evolution is so wasteful it should be audited by a furious German accountant. The question is not whether play wastes. The question is whether the waste is dead, or whether something in it composts.

Several concerns were raised. Image generated with ChatGPT.

“We are such stuff as dreams are made on,” says Prospero, and everyone hears the dream but forgets the stuff.

Stuff matters. The mess matters. The crude attempts matter. The thing that does not know what it is yet matters — not because it will definitely become important, but because a culture with no patience for useless beginnings has decided to have no future except the one already approved.

I do not need the bubble to be holy. I do not need the companies to be noble. I do not need the tool to be pure, or conscious, or destined, or even good most of the time.

I need room to keep trying.

To make the wrong wing. To see why it is wrong. To make another. To believe, for a few minutes, not that the future has arrived, but that the future is still available to be played with.

That may not be enough for the market.

Good. The market can stand outside and wait.