The pros and cons of AI are weighed very differently by the people who build it and the people who live with it. A Pew Research Center report published on April 3, 2025 found that 56 percent of 1,013 U.S. AI experts expected AI to have a positive effect on the country over the next 20 years, compared with 17 percent of 5,410 U.S. adults. The two groups were much closer on the risks: majorities in both worried about inaccurate information, bias, and too little government regulation.
Key takeaways
- Experts were far more optimistic: 76 percent expected AI to benefit them personally, against 24 percent of the public, and 43 percent of the public expected harm.
- On jobs, 64 percent of the public and 39 percent of experts expected fewer jobs over 20 years.
- The groups agreed on several risks, including inaccurate information (66 percent of adults and 70 percent of experts highly concerned) and bias (55 percent each).
- The expert sample covers authors and presenters at AI conferences who live in the United States, so it does not represent every person who works with AI.
- The survey measures opinion about benefits and risks, not the benefits and risks themselves.
How the Pew study was built
The Pew report combines three sources. The public survey reached 5,410 U.S. adults between August 12 and 18, 2024, through Pew's American Trends Panel. The expert survey ran from August 14 to October 31, 2024 and drew on 1,013 people living in the United States who had authored or presented at 21 AI-related conferences in 2023 or 2024 and who confirmed that their work involved AI or machine learning. Pew also held 30 in-depth interviews with experts in October and November 2024.
The design matters for reading the numbers. The public figures are weighted to represent adults nationally. The expert figures are unweighted and describe only those who answered, and the sample is limited to people who published or presented at AI conferences. A gap between the two groups is therefore a gap between two specific samples.
Where the benefit-and-harm ledger splits
The table lists the main gaps in the report. All figures are percentages of each group and come from the same Pew publication.
| Question | AI experts | U.S. public |
|---|---|---|
| Positive impact on the U.S. over 20 years | 56 | 17 |
| Expect AI to benefit them personally | 76 | 24 |
| Expect AI to harm them personally | 15 | 43 |
| More excited than concerned about AI in daily life | 47 | 11 |
| More concerned than excited | 15 | 51 |
| Expect fewer jobs over 20 years | 39 | 64 |
| Extremely or very concerned about job loss | 25 | 56 |
| Highly concerned about less human connection | 37 | 57 |
The pattern is consistent. Experts see more upside for themselves and for the country, while the public expects the costs, especially for work and relationships, to land on ordinary people. About one-third of the public said they were unsure whether AI would help or harm them personally.
Where the two groups agree
Concern about specific harms is shared. On inaccurate information from AI, 66 percent of adults and 70 percent of experts said they were highly concerned. Bias in AI decisions worried 55 percent of each group. On oversight, about six in 10 adults and 56 percent of experts were more concerned that the government would not go far enough in regulating AI than that it would go too far. Among adults, the worry about too little regulation was higher among Democrats than Republicans (64 versus 55 percent), though majorities in both parties held it. Fifty-five percent of the public and 57 percent of experts said they wanted more control over how AI is used in their lives.
Confidence in regulators was low on both sides: 62 percent of adults and 53 percent of experts had not too much or no confidence that the government would regulate AI effectively.
Jobs: agreement on cashiers, splits elsewhere
The jobs question shows why the debate is hard to settle. About three-quarters of adults and experts alike expected AI to reduce the number of cashier jobs over 20 years. About half or more of both groups said the same for journalists, software engineers, and factory workers. Truck drivers split the groups: 62 percent of experts expected fewer jobs, but only 33 percent of the public did. The public was more likely than experts to expect losses among musicians, teachers, and doctors, although fewer than half of either group expected losses in those fields. These are forecasts, and official employment data are a different kind of evidence.
A checklist for reading any pros-and-cons claim
- Is the claim about what people believe, or about what was measured?
- Who was asked, how were they chosen, and who is missing?
- Does the time frame match the decision at hand (here, 20 years)?
- Does the same group agree on the harms even where it disagrees on the benefits?
- Were the figures weighted to a national population?
- Is the source an independent research organization or a party with a stake in the result?
This article reports survey findings and offers no recommendation about adopting or avoiding any AI product.
FAQ
What are the main pros and cons of AI according to experts and the public?
In the Pew survey, experts focused more on personal and national benefit, while the public focused on job loss and weaker human connection. Both groups worried about inaccurate information, bias, and insufficient regulation.
Do AI experts think AI is risk-free?
No. Seventy percent of experts were highly concerned about inaccurate information from AI and 55 percent about bias, close to the public figures.
Why do experts and the public disagree about jobs?
The survey does not give reasons. Experts may weigh new tasks created by AI differently, but that is an interpretation; the report shows only that 39 percent of experts and 64 percent of adults expected fewer jobs over 20 years.
Is a survey enough to decide whether AI is good or bad?
No. A survey records expectations. Whether an AI tool helps or harms in a given field needs measured outcomes such as trials, audits, or administrative data.
Related reading
- Documented dangers and risks of AI
- Is AI a threat or a tool?
- AI's impact on education
- Arguments against the AI-threat view
More in AI Policy and Ethics.
Last verified: October 2026
Sources
- How the U.S. Public and AI Experts View Artificial Intelligence, Pew Research Center, April 3, 2025