The day after I published a piece arguing that AI doesn’t need to be conscious to acquire power, the President of the United States renamed it.
At the U.N. General Assembly, he said the word “artificial” makes AI “sound fake and it is not fake,” and that U.S. documents would use the term “super” instead. That same afternoon, one State Department bureau was told to change every reference to “Artificial Intelligence” to “Super Intelligence” and use “SI” from then on.
My first instinct was to call this another fight over semantics: arguing over the label while the behavior underneath goes unexamined. But there’s a story from an entirely different industry that changed my mind, and it’s barely a year old.
In 2019, the Insurance Institute for Highway Safety ran a simple experiment. It asked more than 2,000 drivers about five driver-assistance systems. Participants were given only the names of the systems. They weren’t told which brand made them and got no other information. All five systems did roughly the same job. None could reliably handle lane-keeping and speed control in every situation, and all of them required an attentive driver.
Only the names differed. One was called Autopilot.
When asked whether it would be safe to take their hands off the wheel, 48 percent of people asked about Autopilot said yes, compared with 33 percent or fewer for the other systems. More of them also thought it would be safe to read a book, talk on the phone or text. Six percent thought a nap would be fine, twice the rate for the other names.
Nothing about the technology had changed. A single word had changed what people were willing to hand over.
Real users showed the same pattern. In a 2022 IIHS study, 42 percent of Autopilot users said they were comfortable treating their vehicles as fully self-driving, against 12 percent of users of Nissan’s ProPilot Assist. The institute’s explanation was blunt. The name Autopilot implies the system is more capable than it is, while “ProPilot Assist” suggests a helper rather than a replacement for the driver. Nissan owners were the least likely to overestimate what their system could do.
California eventually took the name to court. In December 2025, the state DMV adopted a judge’s finding that Tesla’s use of “Autopilot” and “Full Self-Driving Capability” was misleading and violated state law. The judge reasoned that a reasonable consumer would likely believe a car with Full Self-Driving Capability could travel safely without a driver’s constant, undivided attention, and that this belief was wrong both technically and legally. Faced with a possible license suspension, Tesla stopped using the term “Autopilot” in its California marketing and relabeled its premium system “Full Self-Driving (Supervised).”
Consider the timing. In February, a state government made a company downgrade the name of an automated system because the bigger name was changing how people behaved behind the wheel. Seven months later, the federal government upgraded the name of a far more consequential automated technology, by decree, on the world’s biggest stage.
So are we fighting over semantics again? Partly. But the Autopilot story shows why “semantics” is the wrong word to dismiss it with.
A name isn’t a description of behavior. It’s an instruction for behavior. People don’t read a label and then carefully work out how much to trust the thing. The label sets the default level of trust before any thinking happens, and most of what people do happens at the default.
That’s what I was getting at in my last piece. We’ve built professions, organizations and habits around signals that used to be reliable because they were hard to fake: fluency, confidence, memory, credentials, titles. A name that claims superiority is one more of those signals, and it may be the most powerful, because it arrives before any experience with the system and shapes how that experience is interpreted.
“Artificial” was never a flattering word, and I understand why people dislike it. But it did what “Assist” did for Nissan. It kept a little distance between the human and the system and quietly reminded us that we were still the driver. “Super” does what “Autopilot” did. It tells people they can take their hands off the wheel.
Nobody will consciously reason, “The government says it’s super, so I’ll stop checking its work.” That’s not how it works. It happens the way the Autopilot drivers let go of the wheel: gradually, without a decision, because the name made letting go feel reasonable.
In fairness, the case for the rename has a point. What these systems produce is not fake. Driver-assistance systems also have real safety benefits, and nobody sensible argues for abandoning them. The problem was never the capability. It was the gap between what the name promised and the level of supervision the system still needed.
That gap is exactly where AI sits now. These systems are remarkably capable and still need a human paying attention: to check the analysis, question the recommendation, and notice when the output is confident and wrong. The organizations that do well with AI will be the ones whose people keep their hands on the wheel for the right decisions, not because they distrust the tool but because they understand what it is.
A name can help with that or make it harder.
So here’s my answer to whether we’re once again fighting over semantics instead of behavior.
We’re fighting over semantics because semantics drives behavior. The Autopilot case settled that, in a survey, a DMV ruling and a rebrand. The mistake is thinking the naming debate is about respect: whether the technology deserves a more impressive title. It isn’t. It’s about what the title teaches millions of people to do.
California already reached its conclusion about a car that kept its driver in the loop only on paper. It made the name tell the truth, so that “(Supervised)” is now part of the product.
The rest of us may soon face the same question about technology far more powerful than a car. The question isn’t whether AI is artificial or super. It’s whether the word we choose keeps humans supervising it or invites them to stop.
