newsfilter.io
Interview, Fireside Chat

The American Public Kinda Hates AI. Why Exactly? | Dr Yam, Pew Research Center

80,000 HoursDr Yam, Eileen Yam, Rob Wiblin
Watch on YouTube →

Public vs. Expert Sentiment Gaps

  • Experts are significantly more optimistic than the general public regarding AI's impact on productivity: 74% of experts predict increased productivity versus only 17% of the general public.
  • A 73-point disparity exists on job quality: 73% of experts foresee a positive impact on how people do their jobs, compared to 23% of the public.
  • Economic outlooks diverge sharply, with 69% of experts predicting a positive economic effect versus 21% of the general public.
  • Personal benefit expectations show a 52-point gap, where 76% of experts believe AI will benefit them personally compared to 24% of the public.
  • Experts are far more willing to trust AI with critical decisions: 51% of experts express trust compared to only 13% of the public.

Public Concerns and Cultural Context

  • 50% of U.S. adults are more concerned than excited about increased AI use, an increase from 37% in 2021.
  • 61% of the public desire more control over how AI is used in their lives.
  • The primary fear cited by 27% of respondents regarding AI risks is the erosion of human abilities and social connections.
  • Open-ended responses frequently invoked cultural touchstones like Terminator and 2001: A Space Odyssey, indicating a pre-existing cultural script for AI danger.
  • 43% of the general public believes AI is more likely to harm them than benefit them, whereas 76% of experts believe the opposite.
  • 36% of the public believe the media describes AI's impact as smaller than it should be, while another 36% feel it is described correctly; only 21% believe it is overhyped.
  • 33% of the general public remain "not sure" if AI will benefit or harm them personally, compared to a much smaller "unsure" rate among experts.

Specific Societal Applications and Restrictions

  • 66% of the public believe AI should play no role in judging whether two people could fall in love.
  • 73% say AI should play no role in advising people about faith in God.
  • 60% say AI should play no role in decisions regarding how to govern the country.
  • 47% say AI should play no role in selecting members of a jury.
  • 36% believe AI should play no role in providing mental health support, though 46% support at least some role.
  • Conversely, 74% accept AI playing a significant role in weather forecasting, with 88% of postgraduates accepting it versus 62% of those with a high school degree.
  • 70% of the public support AI usage in searching for financial crimes and identifying fraud.
  • 60% of the public say AI should play no role in matchmaking, yet 56% of the public is highly concerned about AI leading to job loss.
  • Experts and the public agree that AI should not have a positive impact in politics and journalism.

Demographic and Attitudinal Differences

  • Gender Gap: Women are more wary than men; only 18% of women expect to personally benefit from AI compared to 31% of men.
  • Expert Gender Gap: Among experts, 63% of men see a positive future impact versus 36% of women, a gap that is more pronounced than in the general public.
  • Age Gap: Younger adults (under 30) are more pessimistic about AI's effect on human skills, with 61% believing it will worsen creative thinking versus 42% of those over 65.
  • Race: Asian American respondents (English-speaking in the sample) were more receptive and enthusiastic about AI's positive societal impact compared to other racial and ethnic groups.
  • Education: Those with higher education levels are more accepting of AI in specific domains like weather forecasting, though overall concern levels by education are not starkly polarized.
  • Political Affiliation: AI remains largely non-partisan at this stage, with both Republicans and Democrats expressing skepticism toward government and industry regulation.

Perception of Control and Misinformation

  • 70% of people say it is important to detect when content is created by AI, but only 50% feel confident they could actually spot it.
  • 18% of those perceiving high AI risks cited inaccurate information as a primary concern.
  • 60–70% of both the public and experts identify AI-generated misinformation as a significant concern.
  • Experts and the public both express low confidence in the ability of either industry or government to effectively self-regulate or regulate AI, respectively.

Future Outlook and Methodology

  • Pew Research Center surveyed over 5,000 U.S. adults across three waves and 1,000 AI ecosystem experts (recruited via conference attendee lists, not random sampling).
  • The study found that while 34% of U.S. adults have used ChatGPT, experts overestimate public engagement, with 70% of experts believing the public interacts with AI daily.
  • Experts anticipate a future of recursively self-improving AI that could displace most office workers within five years, a vision the public finds anxiety-inducing.
  • Future surveys will track longitudinal changes in work, healthcare, and education sectors, as well as evolving public sentiment on guardrails and ethics.
  • There is no consensus on whether AI is currently being overhyped or understated; the public is split 36/36/21 between "understated," "about right," and "overhyped."