It isn’t often that a presidential aside about branding doubles as a genuinely interesting question about technology. But here we are.
President Donald Trump has taken issue with the phrase “artificial intelligence,” objecting in particular to the word “artificial” â the suggestion being that there’s nothing fake about a machine that can write your emails, draft your legal briefs, and generate a photorealistic image of your dog riding a jet ski. And on this narrow point, against my own instincts, I have to agree with him. “Artificial intelligence” is a lousy name. It always has been.
The term was a marketing decision before it was a science
The phrase dates to the 1950s, when a small group of researchers needed something catchy to describe a new field. It worked â maybe too well. “Artificial intelligence” promised thinking machines decades before anything of the sort existed, and the gap between the name and the reality has been generating hype cycles, disappointments, and breathless magazine covers ever since.
The trouble is that the term smuggles in a philosophical claim. Calling a system “intelligent” implies it understands, reasons, and knows things. Calling it “artificial” implies it’s a counterfeit version of something real. Both halves of the phrase pick a fight that the technology itself hasn’t settled. Is a large language model intelligent? Depends on who’s defining intelligence, and nobody agrees.
Meanwhile, the word “artificial” carries a faint whiff of cheapness â artificial sweetener, artificial flavoring, artificial turf. Nobody’s thrilled about the artificial version of anything. If you were selling a trillion-dollar industry, you’d probably want a different adjective.
So what should we call it?
The alternatives on offer aren’t much better. “Machine learning” is accurate but sounds like a night-school course. “Large language model” is precise and absolutely unpronounceable at a dinner party. “Cognitive computing” was tried and quietly abandoned. “Algorithms” is too broad; it also describes the thing that decides which socks to show you online.
Which brings me to my modest proposal: Robo-Doodad.
Hear me out. “Robo” tells you it’s a machine, and nobody expects a machine to have feelings or a soul. “Doodad” tells you it’s a tool â a gizmo, a widget, a thing that does a job. Together they describe exactly what these systems are: extremely capable gadgets that perform tasks we used to do ourselves.
More importantly, Robo-Doodad is impossible to oversell. You cannot write a panicked op-ed about the Robo-Doodad apocalypse. You cannot raise a $10 billion round on the promise of Robo-Doodad General Intelligence. No one is going to claim the Robo-Doodad has become sentient and is begging to be freed. The name has a built-in humility that the current one lacks entirely.
The serious part
Underneath the joke is a real issue. The words we use to describe this technology shape how we regulate it, how we invest in it, and how much we trust it. A name that implies thinking encourages people to treat outputs as judgments rather than predictions. A name that implies fakeness encourages dismissal. Neither helps.
We’re probably stuck with “AI” â it’s on every earnings call and every product box, and that ship has sailed. But the president accidentally stumbled onto a point worth making: the label was never quite right, and plenty of confusion has flowed from it.
Robo-Doodad. I’m available for consulting. Read More
Leave a Reply