Only one in five Americans expects AI to make people more productive, and that gap is where operators earn their standing

ABC News reported on Wednesday that 57% of Americans say artificial intelligence is bad for society. The finding comes from an ABC News/Washington Post/Ipsos poll of 2,545 adults conducted from 9 to 13 October. Two-thirds oppose new data centres. The figures that matter most for an operator sit further down. 65% see a high risk of AI eliminating many jobs. 13% think it has a big chance of creating many new ones. Only 20% think it has a big chance of making people more productive.

The productivity gap is the signal

Beneath the headline, the risk and benefit tables show a credibility deficit. Nearly twice as many respondents — 38% — told Ipsos there is "little or no chance" of AI making people more productive as said there was a big chance. Only 34% see a big chance of everyday tasks getting easier.

Your colleagues and managers come from that population. They have heard the promise. Most of them have not seen it delivered.

The personal numbers make the point sharper. Americans split 50% to 49% on whether AI is good for them personally, and the strength of feeling sits with the pessimists: 15% say extremely bad, 3% extremely good. Younger respondents are the most negative, with 65% of under-30s saying AI is bad for society.

When a room expects a tool to fail, the person who shows it working on a real problem, with the numbers attached, carries unusual weight. Claims about AI are cheap, and audiences discount them. Demonstrated results are scarce.

Judgement is the asset people fear losing

For operators, the most useful line in the poll is this: 62% of Americans see a high risk of people losing the ability to make their own decisions without AI. Among under-50s the figure is 67%. Majorities of Democrats (68%), independents (63%) and Republicans (55%) agree.

That worry points to the layer of work that holds its value. AI does the routine work. You do the thinking. Respondents fear people will hand over the thinking as well. A professional who visibly keeps it, who can explain why a recommendation is right, overrule a model when it is wrong and own the call, is answering the question most respondents are already asking.

Two-thirds also say the federal government is not doing enough to regulate AI, and 49% trust neither party to handle it. Regulators will act in some form. Organisations will then be asked who checked the output and who signed off. Accountability becomes a job in its own right, and someone has to hold it.

Your Next Move

1. Turn one AI use into one number. Pick a task you already run with AI. Record the time saved, the error rate or the turnaround, before and after. When 38% of the public expect little or no productivity gain, a measured result from your own desk carries further than any vendor deck.
2. Keep your judgement visible. For every AI-assisted piece of work you hand over this month, add two lines: what you checked and what you changed. That record shows the decision stayed with you.
3. Check your own role against the job fear. The 65% figure measures worry about job losses in general. To see what it means for your role, use the afternoon method for sorting a week's tasks by what AI can absorb, then build on what is left.