Americans have rendered a verdict on artificial intelligence, and it isn't kind. In the ABC News/Washington Post/Ipsos poll published Oct. 21, 57 percent said AI is bad for society, and 65 percent see a high risk that it will eliminate many jobs. About two-thirds oppose building new data centers, 72 percent if the site is in their own community, and two-thirds disapprove of President Trump's handling of the technology.
Listen to the debate and it's easy to see why. The best case on offer is that AI takes your job. The worst case is that it kills you first. At public hearings over proposed data centers, the only people who speak in favor are the utility and the AI company. Everyone else shows up against, mostly over power and water.
Artificial intelligence has opponents in every district and no constituency. That isn't because the case against it is stronger. It's because the case for it has never been made in terms the average person can picture. This election day, voters are weighing only the costs.
The reason is older than AI. Stand on a Manhattan street corner in 1900 and watch the first automobiles rattle past. Name the jobs about to disappear: teamsters, stable hands, the sweepers behind a hundred thousand horses. Easy. Now name the jobs the car will create. Silence. Nobody says gas station or motel or traffic engineer, because those words don't exist yet. Losses have names. Gains don't, and a mind that can't picture something files it under "probably won't happen."
I see this gap every week. I train people to use AI at work, and before each session I survey the participants — more than 7,000 so far, from nuclear contractors to national park rangers to the NBA. These are motivated people. Sixty-nine percent already use AI daily, and fewer than 4 percent doubt it can help them. Yet 71 percent say the same thing holds them back: "I know that it could be helpful but don't know how to apply it to my job effectively." When the survey asks for ideas about how they might use it, half leave the box blank, even among the daily users.
These aren't skeptics. They're believers with no mental model, people who have the tool open and can't picture what to hand it. If that's true of people who sought out training, voters can be excused for not picturing it either.
So here is the gain, named: help. Executives, surgeons and senators have always had someone to craft the deck, prep the patient and draft the brief. For everyone else, doing the work of three people meant paying two more, so we did without and muddled through. That help now costs less than a streaming subscription, and most of us haven't noticed. I still catch myself writing the first draft of everything, out of habit, with a capable assistant sitting idle one tab over.
Sometimes the help is more than convenience. A mother in one of my training groups had spent a decade taking her bedridden daughter to specialists. She uploaded a thousand pages of medical records and asked an AI for seven theories. The first was a rare condition no one had tested for. Her doctor ordered the test, and the hypothesis proved right. For ten years she had no one to ask. Now she did.
None of this settles the policy fights. The power draw is real, the water is real, and voters are right to press on both. But a serious fight needs two sides, and right now one of them is empty — not because its members don't exist, but because they don't know who they are.
That side can't be recruited by argument. Nobody has ever been talked into wanting help they've never had. It gets built by experience, one person at a time, and the experience is cheaper and closer than the debate suggests.
So, before you vote on AI — on a ballot or at the next zoning hearing — borrow an assistant for a week. Take the insurance renewal you've been dreading or the lab results you don't understand, strip out your personal details, paste them into any of the major models and ask it to explain what you're looking at. Then ask it to draft your reply. Then ask it what you should have asked. You'll be surprised by how much it gets right, and you'll notice what it gets wrong. That's the point: you'll have a judgment formed by use rather than by headlines.
You may still vote against the data center down the road. But you'll be voting on the whole ledger, and the half you couldn't see is a browser tab away.
Jeremy Utley is the author of the forthcoming book The Human Advantage: What AI Can't Do Without You (Hay House, Nov. 10). He has taught innovation at Stanford since 2009 and now teaches through Stanford Online and Harvard's Division of Continuing Studies. He co-hosts the AI-focused tech podcast "Beyond the Prompt."