As one folk theory among AI-watchers goes, all of the hype about it taking jobs, obliterating shared notions of truth, and then ending the world, is actually good for the industry: it amounts to free advertising for the power of the technology. But all of that doom-and-gloom (and the free publicity) has an important downside. September has brought a bumper crop of polling results on artificial intelligence, and they show public views of the technology are dim and getting dimmer. A DFD review of eight polls released in the last month shows people aren’t just worried about their jobs, but also about the technology’s effects on democracy, and they want the government to intervene. Americans are not psyched about AI. Nearly 40 percent of those polled believe it does more harm than good, compared to 10 percent who think it does more good, according to a Gallup survey released this month. A lot of that comes down to economic concerns. Gallup finds that Americans overwhelmingly believe AI will be a job-killer, and an American Psychological Association survey finds that nearly two in five American workers are worried that AI is coming for their personal job duties. However, concerns extend to the tech’s effects on democracy. In Europe, 70 percent of British and German citizens who said they understand AI and deep fakes are concerned about AI’s effects on elections, according to a poll released last week by the nonprofit Luminate foundation. Meanwhile, more than half of Americans said misinformation spread with AI will definitely or probably affect the outcome of next year’s presidential election, and more than a third said they expect AI will decrease their confidence in the outcome of U.S. elections, according to a poll released this month by Axios and Morning Consult. In important respects, the public’s views of AI are getting worse as exposure to it increases. A poll released last week by the survey firm Harris and the Mitre Corporation, a nonprofit that supports federal agencies, found that 39 percent of Americans believe AI is “safe and secure.” That’s down from 48 percent who agreed with that statement when the organizations conducted a poll last November, on the eve of ChatGPT’s release. A Pew survey released late last month found that 52 percent of Americans are “more concerned than excited” about the use of AI in daily life, compared to 10 percent who are more excited. Those numbers are significantly worse than the responses Pew received to that question in 2022, which in turn were marginally worse than the responses in 2021. And these views aren’t just getting worse in the ways that AI companies might like. As the public has gotten a taste of the growing power of generative AI, the Harris-Mitre poll found growing skepticism that AI is ready for primetime. The percentage of Americans who agree that “AI technology is advanced enough for use in mission-critical applications in defense and national security,” declined 8 points to 46 percent between this poll and last November’s Americans want government intervention. The Gallup survey, which was conducted in May and released this month, found that four out of five Americans have little or no trust in businesses to use AI responsibly. The Harris-Mitre poll found that support for consumer protection regulations has grown by three points since last November, hitting 85 percent. The Axios-Morning Consult poll found that a quarter of Americans believe a whole new federal agency should regulate AI, compared to 14 percent who believe regulation is best left to the private sector. The AI Policy Institute — a think tank founded this summer to push the government to check the grandest ambitions of AI companies — released new polling last week, conducted by YouGov, that found a majority of Americans support regulation to prevent the emergence of artificial superintelligence. That builds on polling the think tank released last month that found a majority of Americans support having a federal agency regulate the technology. Despite all the bad vibes for AI, one batch of results validated the theory that alarming coverage about AI is good PR for the field. Last week, online course provider EdX released a survey of 800 C-suite executives that found the group is relatively jazzed about the technology. And it’s not necessarily because they believe their jobs are immune from automation. The survey’s most arresting finding: 49 percent of CEOs want AI to take over most or all of the CEO job.
| 
