In 2024, 67% of respondents said they had a good understanding of what AI is — the same as 2023, up from 64% in 2022. But only 52% said they know which products and services actually use AI, a number that has barely moved in three years. The share who see AI products and services as more beneficial than harmful rose from 52% in 2022 to 54% in 2023 to 55% in 2024. Modest gains, spread very unevenly.

The most revealing pair of numbers sits inside that survey. Asked whether AI products and services have profoundly changed their daily life over the past three to five years, 50% agreed — a figure that has moved one point since 2022. Asked whether AI will profoundly change their daily life over the next three to five years, 66% agreed, up from 60% in 2022. The public consistently reports that the transformation has not happened to them yet and is about to. That gap has widened, not closed, across three years of surveys.

Trust fell on every measure that has one

• Trust that companies using AI will protect personal data fell from 50% in 2023 to 47% in 2024 — a three-point drop.
• Trust that AI will not discriminate or show bias toward any group fell from 56% to 54%. Notably, respondents trust AI on this more than they trust people: only 45% said they trust people not to discriminate.
• Across all questions the average positive response fell 0.6% year over year, driven almost entirely by those two trust measures.
• The share saying AI products and services make them nervous stands at 50%, down slightly from 52% in 2023 but up sharply from 39% in 2022.

Where you live still decides how you feel

The regional split first reported in the 2023 AI Index has not closed. In Asia and Latin America a large majority sees AI as net beneficial: 83% in China, 80% in Indonesia, 77% in Thailand, 70% in Mexico, 62% in India. In Europe and the Anglosphere only a minority agrees: 46% in Great Britain, 44% in Australia, 40% in Canada, 39% in the United States, 36% in the Netherlands. China leads on nearly every positive measure — 81% know which products use AI, 80% are excited by them, 76% trust AI not to discriminate, and 86% expect AI to profoundly change their daily life in the next three to five years. The equivalent American figures are 58% expecting profound change and 34% reporting excitement.

The skeptics moved the most

Among the 26 nations surveyed in both 2022 and 2024, 18 saw an increase in the share believing AI offers more benefits than drawbacks — and the biggest gains came from the countries that started lowest. In 2022 the least optimistic were France (31%), Canada (32%), the United States (35%), Germany (37%), Australia (37%), and Great Britain (38%). By 2024 optimism had grown in France by 10 points, Germany by 10, Great Britain by 8, Canada by 8, and the United States by 4. Belief that AI will profoundly change daily life rose 6 points globally, with every country except India, Malaysia, and Poland increasing — the largest jumps in Canada (17 points) and Germany (15).

Two countries moved backward. Brazil and Malaysia saw the sharpest average declines in awareness, trust, and excitement, both led by falling trust that AI companies will protect personal data. South Africa and Ireland moved the other way; Ireland's gain appears driven by lived experience, since it reported the largest increase of any country in respondents saying their daily lives have been profoundly changed by AI products and services. Anxiety maps onto the same geography: the UK, US, Canada, Australia, and New Zealand report the highest nervousness and lowest excitement, while China, South Korea, and Indonesia report the reverse — with Japan the standing exception among Asian countries.

Broken out, 21% think it very likely and 39% somewhat likely that AI will change how they work; for job replacement the figures are 11% very likely and 25% somewhat likely, with 56% saying it is not very or not at all likely. Compared against 2023 — when 57% expected their work to change and 36% expected replacement — sentiment is essentially flat. The fear of being replaced did not grow at all in a year when the technology visibly did.

Year-over-year comparison on this question needs a caveat: the 2023 survey did not separate 'very likely' from 'somewhat likely,' so the 2024 categories have to be aggregated to compare. Done that way, the picture is stable rather than escalating. What the breakdown does show is how soft the majority is — most of the 60% who expect their work to change sit in the 'somewhat likely' band, and on replacement the largest single response is 'not very likely' at 33%, with another 23% saying not at all likely.

Every generation moved, none flipped

Gen Z remains most convinced their work will change (67% in 2024, up from 66%), followed by millennials (64%, up from 61%), Gen X (55%, up from 53%), and baby boomers (49%, up from 46%). The 18-point spread between Gen Z and boomers is large, but the direction is uniform: every generation is more convinced than it was a year ago, and the largest single-year gains came from millennials and boomers — a sign of awareness spreading across, rather than concentrating in, age cohorts.

Convenience yes, prosperity no

• 55% believe AI will reduce the time it takes to get things done, and 51% expect better entertainment options. Those are the only two majorities in the whole set.
• Confidence drops on everything that touches livelihood: 38% think AI will improve health, 37% their own job, 36% their national economy, and just 31% the job market.
• The economic split is regional again. 72% in China and 54% in Indonesia expect AI to improve their economy; fewer than 25% do in the Netherlands, the United States, Belgium, Sweden, and Canada.
• On health, Mexico is the most optimistic at 56% against a global average of 38%; Japan is the least, at 19%.
• The answers correlate tightly. Countries expecting AI to help the job market also expect it to help their own jobs — Poland at 17% and 21% respectively, China at 44% and 62%. Countries that expect AI to save them time expect it to improve their individual jobs too.

Support for government regulation of AI jumped from 55.7% in 2022 to 73.7% in 2023. The partisan gap is real but the movement is bipartisan: Democratic support went from 74.6% to 84.4%, Republican support from 42.7% to 67.9% — a 25-point swing, the largest single shift in the study. Across both waves combined, 79.2% of Democrats and 55.5% of Republicans agreed AI should be regulated.

The two-wave design is what makes this study useful. Wave one ran in 2022, before ChatGPT; wave two in 2023, after. Because the same population was asked the same questions on either side of that release, the study isolates what a single consumer product did to the regulatory instincts of officials who had mostly never thought about AI. The answer is 18 percentage points of movement on whether it should be regulated at all, concentrated among the group that started most opposed.

Regulate, retrain — but do not redistribute

Asked which specific policies would be beneficial for 2025 to 2050, local officials converged on rules and training and split sharply on money. Stricter data privacy regulation drew 80.4% support, retraining programs for the unemployed 76.2%, and AI deployment regulations 72.5%. Below that the coalition thins: stronger antitrust 57.7%, parole and sentencing AI regulations 54.7%, bias audits for hiring and promotion AI 51.7%, a stronger social safety net 46.4%, semiconductor and AI hardware subsidies 44.4%, a robot tax 42.4%, immigration reform for AI developers 39.1%. At the bottom sit a law enforcement facial recognition ban (34.2%), wage subsidies to offset wage declines (33.9%), and universal basic income — supported by 24.6% and opposed outright by 58.3%.

Nobody thinks it is their problem yet

• Only 34.3% of local officials believe they will need to make AI policy decisions within the next few years; 56.5% think they will not. The share expecting to act rose from 32.2% in 2022 to 36.6% in 2023, with Democrats in 2023 highest at 40.5%.
• Only 29.8% feel adequately informed to make AI policy decisions, against 52.3% who do not. Confidence rose only marginally between waves, from 28.5% to 31.3%.
• The combination is the finding: strong and rising demand for AI regulation, a clear list of preferred policies, near-consensus that action is not imminent, and a majority who do not feel equipped to decide. Republicans are slightly more likely than Democrats to feel informed (31.5% vs. 26.8%) while being far less likely to want regulation.