I’m a validation engineer. I’m not a developer. Before AI, I’d never shipped a production app in my life.

In the last year I’ve shipped:

• A playoff fantasy football site that replaced six years of Google Sheets hell
• A sports stats guessing game
• A calculator that tells you how many kids you could beat in a tug of war
• A satirical investor-relations site for my own abandoned projects
• An emergency pants delivery service (yes, really)

None of that existed before because I couldn’t build it. The skill gap was real. AI didn’t “enhance my productivity.” It took me from zero to one on things I was locked out of.

So when I hear “AI is taking from writers, coders, artists,” I hear it, and I also need to say the other half: for a lot of people, it’s handing them the keys to a door they couldn’t open.

The part where people are right

The fear isn’t stupid. People spent years building skills that a model can now approximate in seconds. That hurts. It’s also fair to be pissed when a model was trained on your work without your permission and is now competing with you. I’m not going to pretend that’s nothing.

The part where I’m pushing back anyway

I’m not using AI to mimic anyone. I’m not asking it to write like a living author or paint like a living artist or clone someone’s voice. I’m using it to build things I already wanted to build but didn’t have the technical chops to execute. That’s not theft. That’s a power tool.

“AI is stealing from artists” gets flattened into “AI is bad,” and those aren’t the same sentence. Plagiarism is bad. Deepfakes are bad. Flooding Etsy with AI knockoffs of a real artist’s work is bad. Using Claude to wire up a scoring engine for my fantasy league is… using a tool.

The ATM story is the one people should actually know

Every tech shift comes with a funeral for jobs that don’t actually die. The cleanest example is ATMs.

When ATMs rolled out in the 1970s and accelerated hard through the ’90s, the assumption was obvious: machine does cash, teller is done. The data says the opposite happened, for a long time.

• In 1985, the U.S. had 60,000 ATMs and 485,000 bank tellers. In 2002, there were 352,000 ATMs and 527,000 bank tellers
• The number of tellers required to operate a branch office in the average urban market fell from 20 to 13 between 1988 and 2004
• Urban bank branches increased by 43% during peak ATM proliferation
• Net result: more total teller jobs, not fewer

And the job itself got better. Tellers became part of what banks call the “customer relationship team.” It’s a different sort of skill. Maybe it’s a higher skill. There is some evidence that their wages have gone up. They are hiring more college graduates as bank tellers. The work got more interesting.

Tellers eventually did decline, but not from ATMs. The huge decline in bank teller employment over the last 15-odd years is mainly a story about iPhones and what they made possible. Different technology, twenty years later.

The point isn’t “don’t worry, nobody ever loses.” The point is the first-order prediction — “machine does task, human is gone” — is almost always wrong. The second-order effects are where the real story lives, and they’re usually weirder than anyone guessed.

Every other example tells the same story

• Portrait painters in the 1840s called photography a refuge for hacks and mechanical frauds. Painting didn’t die. It got weirder and better. Impressionism exists partly because photography freed painters from having to be accurate.
• Calculators were going to make kids bad at math. Kids just do different math now.
• Synthesizers were going to end music. They invented entire genres instead.
• Spell check was going to kill literacy. You’re reading this.
• The internet was going to destroy encyclopedias. It did. I don’t miss them.

Each shift came with real losses. Someone got worse off every time. I’m not pretending the pain didn’t happen. But the pie got bigger, the work got different, and in almost every case we ended up with more of the thing, not less.

My actual position

AI is a tool. It’s the best one I’ve ever used. It lets a guy with no CS degree and no design background make things he used to have to beg or pay other people to make for him. For a lot of people who had ideas but no way to execute, it’s the first real shot they’ve had.

Use it, don’t use it, hate it, whatever. I’m going to keep using it to ship weird stuff and write down what happened.

— CT

Read more at Build and Bail