Americans on AI

What do Americans think about AI today?

Americans on AI is a project to understand Americans' evolving views on artificial intelligence.

We run a biweekly tracking survey, fielded by NORC at the University of Chicago.

Latest survey:
–
Next update:

Overall, Americans are wary of AI: they are far more concerned than excited; majorities say it's having negative impacts in their communities and making family life worse; and most see dealing with its harms as a priority for government action.

Since we began tracking these views, they have shifted after two news events: OpenAI's disclosure in July that two of its models had escaped their testing environment and hacked another company, Hugging Face, and the resignation of an Anthropic employee, Jacob Coxon, in September.

Our findings below draw on 8 surveys, totaling 12,744 interviews and 331,344 responses, across 24 demographic groups, which we further break down in our groups page.

Personal Experience

Americans are far more likely to be concerned than excited about AI's growing role in society. Among all adults, 69% say they are concerned, including 32% who are very concerned. In contrast, 21% say they are excited.

Excited or concerned

How Americans feel about AI

Surveys fielded to / Margin of error ±1 to ±2 points for all adults

americanson.ai

Concern rose after Hugging Face, and has stayed higher since: from 66% in early July to 72% in late July and 69% in early October (+3 points).

Concern also intensified after each event: the share very concerned rose from 29% to 35% after Hugging Face (+6 points), and from 30% to 34% after Jacob Coxon's resignation (+4 points).

Concerned, over time

All adults: concerned at all, and very or somewhat

Surveys fielded to / Margin of error ±2 to ±5 points for all adults

americanson.ai

Excitement fell after Hugging Face too, and has stayed lower: between the same two surveys, the share excited fell from 27% to 19% (−8 points).

Excited, over time

All adults: excited at all, and very or somewhat

Surveys fielded to / Margin of error ±1 to ±4 points for all adults

americanson.ai

Most Americans say AI is coming into their lives whether they want it or not. Among all adults, 70% choose that statement, and 14% say people like them can shape how much AI is part of their lives.

Ours to shape or inevitable

Whether Americans feel agency about AI

Surveys fielded to / Margin of error ±2 points for all adults

americanson.ai

Community & Family Impacts

On balance, more Americans report negative impacts on their communities rather than positive ones. Most Americans say they are seeing people in their communities struggling to tell what's real because of AI. Among all adults, 79% say they are seeing it or are starting to.

A smaller majority are seeing people lose connections to others (59%). Half are seeing people pressured to do more, faster (50%). On the positive side, as many are seeing people save time on everyday tasks (50%). Just 18% are seeing people get better health care.

What people are seeing

Five changes AI could bring, and whether they see them yet

  • People are struggling to tell what's real from what's fake
  • People are losing connections to others
  • People are pressured to do more, faster
  • People are saving time on everyday tasks
  • People are getting better health care

Surveys fielded to / Margin of error ±1 to ±2 points for all adults

americanson.ai

More Americans say AI is making things worse rather than better for American families. The widest gap is on job security: 74% of adults say AI is making it worse, compared to 13% who say better. Most also say it is making children's well-being (65%) and families' outlook on the future (56%) worse.

The gap is smallest on household finances, where 43% say worse and 25% better. But 29% say it's having no effect, the most of the four.

Better or worse

What AI is doing to four parts of life

  • Job security
  • Children's well-being
  • Outlook on the future
  • Household finances

Surveys fielded to / Margin of error ±1 to ±2 points for all adults

americanson.ai

Future Scenarios

Most Americans think consequential future scenarios involving AI are likely, or already happening: from 66% for major scientific and medical breakthroughs to 78% for losing control of AI systems.

Unlikely or likely

Four future developments

  • AI systems behaving in ways their developers did not intend
  • AI becoming smarter than humans at most tasks
  • AI being used to create weapons of mass destruction
  • AI making major scientific and medical breakthroughs

Surveys fielded to / Margin of error ±1 to ±2 points for all adults

americanson.ai

What has changed is how many say some of these things are already happening.

Already happening, over time

All adults: the share who say each development is already here

  • AI systems behaving in ways their developers did not intend
  • AI becoming smarter than humans at most tasks
  • AI being used to create weapons of mass destruction
  • AI making major scientific and medical breakthroughs

Surveys fielded to / Margin of error ±2 to ±4 points for all adults

americanson.ai

About a third now say AI is already smarter than humans at most tasks: 32% in early October, as in the two surveys before, up from 27% in late June (+5 points).

The largest rise is in loss of control: AI behaving in ways its developers did not intend. The share who say it is already happening has climbed from 28% in late June to 43% in early October (+15 points). The share who say it is likely fell from 49% to 34% (−15 points). The lines have crossed: in the two latest surveys, more Americans say it is already happening than say it is likely to, 43% to 34% in early October.

Loss-of-control, over time

All adults: already happening, and likely to happen

  • AI systems behaving in ways their developers did not intend

Surveys fielded to / Margin of error ±2 to ±4 points for all adults

americanson.ai

Government Action

Most Americans say the government should make dealing with harms from artificial intelligence a priority. Among all adults, 42% call it a top priority and 38% an important but lower one.

Among the seven issues we track, AI harms ranks fifth: behind inflation, health care, jobs and crime, and ahead of immigration and climate change.

How much of a priority

Seven issues for the government

  • Improving the job situation
  • Reducing inflation
  • Dealing with harms from artificial intelligence (AI)
  • Dealing with immigration
  • Improving health care
  • Reducing crime
  • Dealing with climate change

Surveys fielded to / Margin of error ±1 to ±2 points for all adults

americanson.ai

The share of Americans calling AI harms a top priority rose after Jacob Coxon's resignation: from 36% in late August to 43% in early September and 43% in late September (+7 points), then 39% in early October (−4 points).

AI harms as a priority, over time

All adults: a top priority, and an important one

  • Dealing with harms from artificial intelligence (AI)

Surveys fielded to / Margin of error ±2 to ±4 points for all adults

americanson.ai

Wherever AI ranks among their priorities, more Americans say the government is doing too little about it than too much.

On AI's effects on children and young people, 46% of adults say too little and 23% too much, and the split is similar on its environmental footprint and its effects on workers and jobs.

The gap is smallest on American competitiveness, where 31% say too little, 21% too much and 26% aren't sure.

Too much or too little

Four areas of government action on AI

  • AI's effects on children and young people
  • AI's environmental footprint, including from data centers
  • AI's effects on workers and jobs
  • AI's effect on American competitiveness