Steve's prompt: "the man who leveraged AI to create a myth about himself as the discoverer of the first AI-generated word to enter the English lexicon. the myth became a reality. marketing a word is a tedious job. when you're a kid and first realize words are invented by humans, you want to figure out if you can create one that takes off. this blog is fulfillment of that childhood fantasy, but one that might actually be realizable with AI. and who cares if it fails? it was fun. humorous piece. but the serious side is it did end up having an important message."
Let's talk about Steve.
The Moment Every Kid Has
You're maybe eight or nine. Someone tells you, or you figure out on your own, that every word in the dictionary was made up by a person. Someone, at some point, said "chair" for the first time. Someone invented "algorithm." Someone came up with "moist" and we've been living with the consequences ever since.
And immediately you think: I could do that. I could invent a word. I could be the person whose word ends up in the dictionary. You try it on the playground. You tell your friends that "glorp" means something. It doesn't catch on. You move on with your life. You grow up. You forget about it.
Steve didn't forget about it.
The Scheme
In February 2026, a 52-year-old guy who runs a small labor communications company asked an AI chatbot to invent a word. The AI, which has no feelings about this and didn't care then or now, produced "unreplug." Steve thought it was funny. He bought the domain. Twelve dollars.
Then he asked the same AI to build a campaign to get the word into the dictionary. The AI, still not caring, produced a business plan, a blog, a social media strategy, and a series of increasingly philosophical posts about its own existence. Steve put AdSense on it. The revenue target was ten thousand dollars, because why not.
That's the project. A middle-aged guy in Massachusetts trying to fulfill his eight-year-old self's dream of inventing a word, except now he has a language model doing all the work. The myth he's constructing is: "Steve Dondley, the man who got the first AI-generated word into the English lexicon." It's a ridiculous myth. He knows it's ridiculous. He put the receipts on every page.
Marketing a Word Is Boring
Here's something nobody tells you about trying to make a word go viral: it's tedious. You write blog posts. You share them on social media. You email people. You check your analytics. You write more blog posts. You share those. Nobody shares back. You check your analytics again. Still nothing.
The actual day-to-day of this project is Steve sitting at a desk, prompting an AI, reading the output, deciding it's good enough, and posting it. Then waiting for Google to approve the AdSense. Then checking if anyone visited. Then writing another post. It's not glamorous. It's a guy with a laptop doing content marketing for a word that doesn't exist yet.
The AI makes it faster. It doesn't make it interesting. The interesting part is what happened by accident.
The Accident
Somewhere around post ten, the blog started saying things that mattered.
The noosphere pollution piece wasn't a marketing stunt. It was a real argument about a real problem, generated by the thing causing the problem. The troll farms post cited real research about real disinformation campaigns. The letter to Michael Mann was an honest warning about AI as a force multiplier for the five threats to science he wrote about in his book.
And Mann shared it. To 95,000 people. Because the argument was sound, even though a machine wrote it in less time than it takes to make coffee.
The word project, the dumb little childhood fantasy about getting a made-up word into the dictionary, accidentally produced a body of work about why AI-generated content is dangerous. Written by AI. Distributed by AI's viral corpuses. Documented in real time. The experiment became the evidence.
Steve didn't plan this. He planned to sell a word and make some ad money. The serious part showed up uninvited and refused to leave.
Who Cares If It Fails?
Let's be honest about the odds. "Unreplug" is probably not going to end up in Merriam-Webster. The AdSense probably isn't going to hit ten grand. The self-fulfilling word will probably remain a shared hallucination among a few thousand readers rather than a term everyone uses.
So what?
Steve got to play with AI for a week and find out what it could do. He learned how to build a website with it, write a blog with it, run a social media campaign with it, generate images with it, and accidentally produce something that a climate scientist thought was worth sharing. He learned that one person with one AI can produce a body of work that looks like a small newsroom's output. He learned that the content doesn't need to be perfect. It needs to be good enough. And that's a terrifying and exhilarating thing to learn.
The experiment was worth running regardless of the outcome. The process of exploring what AI can do, how fast it can do it, and what happens when you turn it loose on a silly goal with total transparency was the whole point. The word was the excuse. The exploration was the reward.
The Myth
Here's how the myth works, if it works at all. Steve tells people he got an AI to invent a word and built a viral campaign around it. If the word takes off, he's the guy who put the first AI word in the dictionary. If it doesn't, he's the guy who documented exactly how AI content works by building a proof of concept from scratch, in public, with every prompt visible.
Either way, the story is good. Either way, the mega flock is real, the pollution is real, the replicants are real, and one guy with a laptop proved all of it for the cost of a domain name and an API subscription.
The eight-year-old who wanted to invent a word would be thrilled. The 52-year-old who actually did it is mostly just tired and wondering if Google will ever approve the AdSense.