AI job-loss expectation rises with income and awareness, and that describes your buyer

Pew Research Center published a 37-country survey on September 17 finding that in 34 of those countries, people are more likely to expect artificial intelligence to reduce the number of jobs over the next 20 years than to increase it. In the United States the share expecting fewer jobs is 71%. In Australia and South Korea it is 76% each.

The headline number is not the part that should change anyone's copy. The part that should is who holds the view. Pew reports that job-loss expectation tracks a country's GDP per capita; that across many middle-income countries it also rises with education and household income; and that in around half the sample it rises with how much a person has heard about AI. Rearrange those four variables and you have a working description of a software buyer.

What the survey measures, and what it does not

Pew surveyed 42,151 adults across 36 countries between February 8 and May 13 of this year, with the American figures drawn from two American Trends Panel surveys of 5,119 and 3,488 adults, and groups countries by World Bank income classification. Search Engine Journal's read of the report supplies the medians — 46% across all 37 countries expecting fewer jobs, 9% expecting more, 25% unsure, 55% against 36% between the high- and middle-income groups, and unsure shares of 22% and 34% in those same two groups — plus a GDP-per-capita correlation of 0.60. Pew's own summary pages describe those relationships without printing the figures.

The questions are about AI in the abstract — jobs, inequality, daily life. Nobody was asked whether they would buy anything. Search Engine Journal's Matt G. Southern says the results "don't show how people respond to AI features, products, or advertising," and that limit is real. What the data supplies is the prior: what a buyer already believes when your homepage loads.

Awareness is the variable that should worry you

Pew treats AI awareness as a demographic. In software marketing it behaves more like a qualification filter: the person Pew classes as highly aware — a heavy consumer of AI coverage — is the one who books the demo, starts the trial and writes the internal memo arguing for budget. Pew's findings about that group point in two directions at once.

Across roughly half the markets in the study, that best-informed group was likelier than everyone else to predict job losses. Yet in many of the same markets, it was also likelier to feel optimistic about AI's growing presence in daily life. Germany is the case Pew works through: among the best-informed Germans, 26% lean more concerned than excited; among the least-informed, 41% do.

The better a buyer understands AI, the more they believe it works — and the more certain they are it will cost somebody a job.

Held together, those two findings break the assumption underneath most AI messaging. The standard move treats concern as an information deficit: demonstrate the capability, publish the benchmark, and the worry dissolves. Pew's numbers suggest capability proof lifts both readings simultaneously. Convince someone that the software genuinely does the work and you have also convinced them that the work is being taken from somewhere. Usually somewhere near them.

That last gap is usually read as a maturity difference that will close on its own. The unsure column says otherwise: in the median middle-income country, a third of adults cannot call it either way, half again the high-income rate. Those markets have not decided AI is benign. They have not decided.

The line is moving, and the age split has flipped

The US is the only country where Pew has asked the jobs question before, and the share expecting fewer jobs has risen seven points in two years. On the broader question of concern about AI in daily life, Australia is up four points since last year, Hungary six, the Netherlands seven, Poland eight and Sweden nine — four of them European markets that sit on most B2B expansion plans.

The age data is sharper. Among Americans aged 18 to 34, the share more concerned than excited about AI went from 40% in 2024 to 55% today, and Pew notes that under-35s are now as likely as people over 50 to be primarily concerned, reversing the pattern of five years ago. If your product grows bottom-up, the group most fluent in AI and most convinced it destroys jobs is your daily active user.

What a software marketer should change

• Split the buyer from the displaced. "Replace your SDR team" is a promise to a CFO and a threat to everyone who has to adopt the thing. When both read the same page, write for the adopter and move the headcount arithmetic to the procurement deck.
• Stop shipping one global message. A 55% expectation with 22% undecided and a 36% expectation with 34% undecided are not the same objection. The first market needs a redeployment story; the second needs a plain explanation of what the software does.
• Name the work that disappears, precisely. A vague efficiency claim is an empty field the reader fills with their own org chart.
• Instrument the claim, not the category. Pew measures sentiment toward AI in general; only your win-loss notes say whether "agentic" in your H1 earns anything.