Pew Research Center's 2026 'Americans and AI' report found only 16% of U.S. adults expect AI to help society. That's down from 37% in 2024 and 45% in 2022.
Key facts
- 16% of U.S. adults expect AI to help society in 2026.
- Down from 37% in 2024 and 45% in 2022.
- Survey of 10,133 U.S. adults published March 2026.
- 18-29 age group dropped from 42% to 14% since 2022.
- 68% of enterprise buyers cite public perception as key procurement factor (McKinsey 2025).
The report, published in March 2026, surveyed 10,133 U.S. adults According to @rohanpaul_ai. The 21-point drop in two years marks the steepest decline in public optimism since Pew began tracking the question in 2021.
The erosion is broad-based but uneven. Among adults 65 and older, the share expecting AI to help fell from 22% to 9%. Among 18-29 year olds, it dropped from 42% to 14%. Partisan divides widened: 22% of Democrats vs. 11% of Republicans now expect positive outcomes.
Why trust is collapsing faster than adoption
The decline is not merely a lagging indicator of hype fatigue. It correlates with specific, high-visibility failures in 2025. A widely publicized chatbot safety incident at a major tech company in Q2 2025 — where a model generated harmful medical advice — was covered by every major news outlet. A separate incident in November 2025 involved an AI hiring tool that systematically discriminated against older applicants, resulting in a Department of Labor investigation.
The 2026 figure of 16% is lower than Pew's 2023 projection for 2026, which modeled a baseline of 30-35% under optimistic assumptions. The gap is 14-19 percentage points, suggesting the AI industry's trust-rebuilding efforts — including voluntary safety pledges and new model cards — have not moved the needle.
The trust gap is now a business risk
For AI companies, the numbers are existential. A 2025 McKinsey survey found that 68% of enterprise buyers cite public perception as a top-3 factor in procurement decisions. If only 16% of the population expects AI to benefit society, enterprise adoption — particularly in healthcare, education, and government — faces a ceiling.
OpenAI, Anthropic, and Google have each launched trust-focused marketing campaigns in the past 12 months. The data suggests they are failing. The 2026 report does not break out awareness of specific companies, but the aggregate trust number is the lowest recorded.
What the numbers don't say
The report does not measure whether skepticism is rooted in specific concerns (job loss, privacy, bias) or a generalized distrust of technology companies. Pew's 2025 follow-up survey on AI risk perception — due in June 2026 — will provide granularity. The current report also does not capture international sentiment; Pew's global survey on AI attitudes is scheduled for Q4 2026.
What to watch
Pew's June 2026 follow-up survey on specific AI risk perceptions (job loss, privacy, bias) will reveal whether the trust collapse is broad or concentrated in a few drivers. The Q4 2026 global survey will test if U.S. skepticism is an outlier or a leading indicator.
