AI Is Coming for My Job. Good.
There's an argument about AI that keeps moving the furniture.
It can do production, but humans provide judgment.
If it gets better at judgment, humans provide vision.
If it gets useful at vision, humans provide intent.
At some point we're going to retreat into a cupboard marked EXCLUSIVELY HUMAN VIBES and defend it with a fucking broom.
I understand the impulse. People want to know where they fit. Income matters. Having work disappear underneath you can be frightening.
But I keep getting stuck on the assumption behind the argument: that the interesting destination is a patch of work the machines can't enter.
Why?
If a tool can help with another layer, I have another capability available to use.
I don't immediately need to relocate my entire sense of purpose to the next hill.
The Next Layer
Suppose AI becomes useful at comparing designs.
Good. I can compare more designs.
Suppose it becomes useful at proposing product directions.
Good. I can examine proposals I might not have considered.
Suppose it can help clarify intent by exposing contradictions in what I say I want.
That also sounds useful. I contain contradictions. Several are probably talking right now.
None of this requires pretending the output is correct. I still have to evaluate it, decide what to attempt, and deal with the consequences.
But I don't need to declare today's remaining human task permanently inaccessible to machines in order to enjoy doing it today.
Maybe the boundary moves again. Fine. I'll look at what moved.
“Where will I still be necessary?” puts me on patrol around a shrinking perimeter.
“What can I attempt now?” gives me something to build.
The Tiny Weird Thing
When production gets cheaper, one possible response is to increase scope.
A bigger game. More assets. More dialogue. A map so large that navigating it becomes a second unpaid job.
Fine. Somebody will want that.
But cheaper production can change the other end of the market too.
Imagine a small, peculiar game for a small, peculiar audience. Or a specialist utility that solves one irritating problem for people whose collective purchasing power will never summon a venture capitalist.
Previously, the cost of making it might have exceeded anything its audience could support. Lower that cost enough and the same audience may become sufficient.
The audience didn't get bigger. The thing became possible at a smaller scale.
That's interesting to me.
I don't need every improvement in tools to become a mandatory increase in production values. Sometimes I would like the same modest thing to require less money, less time, and fewer reasons to abandon it.
Let the giant studio build a continent with individually simulated teaspoons.
I might want to make a weird little game that twelve hundred people love.
Or twelve.
The spreadsheet can judge me later.
Cheap to Make Is Only One Part
Of course production isn't the whole bill.
People still have to discover the thing. It has to work. Someone has to maintain it, support it, and decide whether continuing is worthwhile.
A prototype becoming cheap doesn't magically make a business viable.
But reducing one serious barrier still matters. It can make an experiment affordable. It can let a small audience support something that previously needed a much larger one. It can let me find out whether an idea deserves more work before I spend years proving that it didn't.
That's enough to be interested.
I don't need a promise that every weird idea becomes profitable. Most of mine would find that level of responsibility deeply upsetting.
Good for What?
The title doesn't mean I'm cheering for people to lose their income.
It means I welcome the possibility that more of the labor between an idea and a useful result can become easier.
The benefits and the disruption won't distribute themselves fairly by magic. That's a real problem. It deserves more than telling worried people to be enthusiastic.
But preserving a particular chore forever is a strange place to anchor my own response.
If I enjoy doing it manually, I can keep doing it. Enjoyment is a perfectly good reason. I don't need to prove the machine is incapable in order to justify my afternoon.
And when I want the result more than I want the chore, I'll use the available help.
Maybe AI takes another piece of what I currently call my job.
Then I want to know what I can do with that capability.
I have enough ideas waiting around already.
Some of the little bastards might finally get built.