More people expect AI to cut jobs than create them in 34 of 37 countries

Pew's four answer shares add up to 93 ^[22]. TechSpot describes them as a 37-country median: 46% expecting fewer jobs, 13% expecting little difference, 9% expecting more jobs, 25% unsure ^[3]. Each share is the middle country's answer for that one option, and the middle country is not the same country across options, so the four are not slices of one sample. Quoting the 46 against the 9 as a five-to-one margin ^[23] treats two separately computed medians as one population. The US figure sits 25 points above the median on the fewer-jobs option ^[24].

For anyone planning an internal rollout, the within-country breakdowns carry more than the headline does. In many of the middle-income countries surveyed, wealthier and more educated respondents were the ones more likely to expect AI to reduce job numbers over the next two decades, and TechSpot attributes much of that gap to greater uncertainty among people with lower incomes and less education ^[12]. TechSpot also reports that people with high awareness of AI are more concerned about its impact on jobs ^[13]. In a number of countries the most concerned age group is 18 to 34, and concern in the US has grown most in that group ^[10].

The concerned-against-excited split is narrower than the jobs question. Across the 37 countries, a 37% median were more concerned than excited, a 41% median were equally concerned and excited, and Israel is the only country where more respondents were primarily excited ^[14].

Executives are on the same side of the jobs question as the public. A Mercer survey cited by TechSpot found that 99% of the executives questioned expected AI to lead to job cuts within two years ^[15]. That 99% is what executives told Mercer. Meta cut 10% of its workforce and then abandoned plans for a second wave amid disappointing AI results and employee resistance ^[16]. Some employers have rehired workers they replaced with AI after the automation failed to deliver ^[17].

The measured job losses in the same newsletter are smaller than the expectation. The Bay Area tech workforce is about 375,000 people and is down roughly 7% since 2024, according to the analysis The Deep View cites ^[19]. The one-year notice count Bloomberg published works out to about 3.9% of that standing workforce ^[25].

These surveys do not test how an internal announcement moves an employee's expectation. They map the starting position an announcement lands on, and it was already there in March, when Quinnipiac University found 70% of Americans believed AI will reduce job opportunities and 55% believed it would do more harm than good ^[20]. Jason Hiner wrote in The Deep View that Pew data "shows people overwhelmingly expect AI to reduce jobs, leaving Silicon Valley with a credibility gap it still hasn't closed" ^[21].