Pew's 37-country survey puts "fewer jobs" ahead of every other answer in 34 of them

Pew asked people what AI will do to the number of jobs in their country over the next 20 years. In 34 of 37 countries the job-loss answer beat the job-creation answer ^[11]. That answer shows up in your rollout channel anyway, because the person reading the launch note has already given one.

The median shape matters more than the US headline. Across the 37 countries, 46 percent said fewer jobs and 9 percent said more, close to five to one ^[13][21]. Roughly a quarter said they were not sure, which puts the unsure ahead of the optimists by nearly three to one ^[13][23]. Pew reported that in middle-income countries higher earners and the more educated were more likely to predict job loss, and that lower earners and the less formally educated were more likely to answer "not sure" without holding a more positive outlook ^[16].

Fieldwork ran from February 8 to May 13 ^[12]. The Verge notes the survey closed well ahead of the recent apocalyptic warnings ^[7], and cites Anthropic CEO Dario Amodei's warning that AI could eliminate half of entry-level white-collar jobs ^[8]. The US figure of 71 percent was recorded before that, and it was already 7 points above the 64 percent Pew measured two years earlier ^[2][20].

Against that, the measured record is thin, and a team using it to reassure staff is answering a 20-year forecast with one year of one country's data. Gizmodo cites work published in July 2026 by the Stanford Institute for Economic Policy Research finding the overall consequences for the US labor market had been marginal so far, though AI may already be contributing to a harsher environment for recent graduates ^[19].

Movement in sentiment is recent and specific. "The share of adults who say they are more concerned than excited about the technology has gone up since last year in Sweden (+9 percentage points), Poland (+8), the Netherlands (+7), Hungary (+6) and Australia (+4)," Pew wrote ^[5]. On the money question, Pew added: "On balance, people are more likely to say AI will increase rather than decrease the gap between the rich and the poor" ^[9]. In the US, about 46 percent expected AI to worsen inequality and 5 percent expected it to make the country more equitable ^[10].

The two publishers describe the age pattern differently. Gizmodo reports Pew found young adults the most anxious about AI's effect on the job market in both high- and middle-income countries, though not universally ^[6]. The Verge reports that older people are more likely to be more concerned about AI's impact on daily life, and that 18-to-34s are the group most likely to have mixed feelings ^[17]. Those are two questions, and the daily-life one is less lopsided: a global median of 41 percent said they were equally concerned and excited, 37 percent primarily concerned and 13 percent primarily excited ^[18]. That means 54 percent are not primarily concerned about AI in daily life ^[22].

So for Monday, sort on two axes and leave sentiment out of it. First, who turned the feature on: the user, or someone above them. Second, whose step it removes: the user's own, or a colleague's. The box to plan for is the one where an admin enabled it and it removes work someone else was paid to do. In that box I would not read early usage as acceptance, and I would watch whether people come back in week four, whether they finish the same task with it twice, and how long the first useful output takes. These expectations do change. In the US the figure moved 7 points in two years ^[20].