The world does not agree about AI — and 2022 put numbers on the disagreement

Globally, 64% of respondents said they have a good understanding of what artificial intelligence is, 60% expect AI products and services to profoundly change their daily life in the next three to five years, and 60% say those products make their life easier. But only a slight majority, 52%, agree that AI products and services have more benefits than drawbacks — and 39% say they make them nervous.

Two gaps inside that one survey are worth pulling out. The first is between the future and the present: 60% expect AI to profoundly change their daily life in the next three to five years, while 49% say it already has over the past three to five. The second is between claimed knowledge and actual recognition: 64% say they understand what AI is, but only 50% say they know which types of products and services actually use it. On trust the sample splits exactly down the middle — 50% say they trust companies that use AI as much as they trust other companies.

Within a country, every answer moves together

Sentiment is strongly correlated inside each country rather than by question. Chinese respondents are among the most positive on every measure: 87% say AI products make their lives easier, 76% trust companies that use AI as much as other companies, 76% know which products use AI, and only 30% report feeling nervous. American respondents are among the most negative: 41% say AI makes life easier, 35% trust AI companies, 39% know which products use AI, and 52% feel nervous. Japan is the standing oddity — only 42% see more benefits than drawbacks, yet just 20% say AI makes them nervous, the lowest anxiety of the 28 countries. India runs the opposite way: 71% see more benefits than drawbacks and 53% feel nervous, the highest anxiety in the survey.

A 47-point spread between first and last

• China (78%), Saudi Arabia (76%), and India (71%) lead the question of whether AI products and services have more benefits than drawbacks.
• France (31%), Canada (32%), the Netherlands (33%), and the United States (35%) sit at the bottom; the report singles out France and Canada as holding the most negative views of the 28 countries surveyed.
• Emerging economies cluster at the top — Peru 70%, Mexico 65%, Malaysia 65%, Colombia 64%, Chile 63%. Wealthy Western countries cluster at the bottom — Sweden 40%, Belgium 38%, Great Britain 38%, Australia 37%, Germany 37%.
• Expectation follows the same map. 80% in China and Saudi Arabia expect AI to profoundly change their daily life within three to five years, against 46% in the United States and Great Britain and 44% in Canada and Germany.
• So does the sense that it already happened: 73% in China say AI has profoundly changed their daily life over the past three to five years, against 36% in the United States and 30% in Japan and Sweden.

Men score higher than women on every positive statement in the IPSOS set: understanding what AI is (69% vs. 60%), knowing which products use it (55% vs. 46%), believing benefits outweigh drawbacks (55% vs. 49%), saying it makes life easier (62% vs. 58%), expecting it to change their life (63% vs. 57%), and trusting companies that use it (53% vs. 47%). There is exactly one statement where women score higher: AI products make me nervous, 41% against 38%.

Age splits the question in two. Respondents under 35 are the most likely to say AI products make their lives easier (64%, against 62% for 35-to-49 and 54% for 50-to-74), yet they are less likely than the 35-to-49 group to say the benefits outweigh the drawbacks (47% against 53%). Convenience and endorsement are not the same thing. Income, education, and job seniority move together and in one direction: 57% of high-income respondents see more benefits than drawbacks against 50% of low-income; 59% of the most educated against 45% of the least; and among employment groups, senior executives and decision makers top the list at 64%, business owners at 63%, the employed at 55%, and the non-employed at 47%.

A different survey, a blunter question, the same shape

In 2021 Lloyd’s Register Foundation and Gallup asked 125,911 people across 121 countries whether AI will mostly help or mostly harm people in their country over the next 20 years. 39% said mostly help against 28% who said mostly harm — and the gender gap from the IPSOS survey reappears intact: 42% of men said mostly help, against 35% of women. Regionally, Eastern Asia returns 4.4 “mostly help” answers for every “mostly harm,” Northern and Western Europe 1.8, Australia and New Zealand 1.7. Northern America and the Middle East land at exactly 1.0 — one hopeful answer for every fearful one. Eastern Africa (0.4), Southern Africa (0.6), and Northern Africa (0.7) are the most pessimistic regions in the world on this question.

Nobody wants to get in the car

The same poll asked about self-driving cars, and the answer is one of the least ambiguous numbers in the chapter: 65% would not feel safe in a self-driving car, 27% would, and 8% did not know or refused to answer. Pew found the American version of the same instinct — only 26% of Americans think driverless passenger vehicles are a good idea for society, against 44% who call them a bad idea and 29% who are not sure. It is the only one of three specific AI use cases Pew tested where “bad idea” outpolls “good idea.”

45% of Americans report feeling equally concerned and excited about the increased use of AI programs in daily life. 37% are more concerned than excited. Only 18% are more excited than concerned. The largest single group in the country has not made up its mind.

Excitement drops as the task gets more human

Asked about six specific applications, the ordering is almost perfectly mechanical-to-personal. 57% are very or somewhat excited about AI performing household chores, against 19% concerned. Repetitive workplace tasks: 46% excited, 26% concerned. Diagnosing a medical problem: 40% excited, 35% concerned. Handling customer service calls flips the sign — 27% excited against 47% concerned. And at the far end, where the task is judging people rather than doing work, the collapse is total: making important life decisions for people draws 74% concerned against 9% excited, and knowing people’s thoughts and behaviors 75% against 9%.

What people say when you ask them why

• Among Americans who are more concerned than excited, the leading reasons are loss of human jobs (19%), surveillance, hacking, and digital privacy (16%), and lack of human connection or human qualities (12%).
• Lack of oversight and regulation is named by only 2% — tied at the bottom of the list with loss of freedom and human bias coded into AI. The public worry is about consequences, not governance.
• Among those more excited than concerned, 31% point to AI making life and society better, 13% to saving time and increasing efficiency, 10% to it being inevitable progress, and 7% to handling mundane, tedious tasks.
• On specific deployments, 46% of Americans call police use of facial recognition technology a good idea for society against 27% who call it bad, and social media companies using AI to find false information draws 38% good against 31% bad.

Whose experience do Americans think AI was built around?

Pew also asked whose experiences and views are taken into consideration when AI systems are designed. 51% said the experiences of men are well considered, against 36% for women — a 15 percentage point gap. For white adults the figure is 48%; for Asian adults 33%, Black adults 24%, and Hispanic adults 23%. Between 37% and 43% of respondents answered “not sure” on every single row, which is a finding in itself: on the question of who AI is built for, the most common American position is that nobody outside the room can tell.