Americans side-eye AI in new Gallup poll

After months of hand-wringing over whether a fresh AI wave would doom humanity—or at least take over the workplace—it may be unsurprising that Americans aren’t entirely sold on the technology.

A Gallup poll published this week found that four in five US adults don’t trust businesses to use AI responsibly, and three-quarters expect it to take away human jobs. Around 40% of those surveyed thought AI does more harm than good, while 50% were split on the question.

That distrust has yet to rain too hard on businesses’ parade, which have rushed to integrate various forms of generative AI into both enterprise and consumer products. But a growing pile of data showing a sizable level of AI skepticism or disinterest among the public could prove a problem as companies seek to make the tech commonplace.

Noah Giansiracusa, an associate professor of mathematical sciences at Bentley University, which partnered with Gallup on the survey, told Tech Brew he was surprised to see the extent to which concerns typically confined to tech-critic circles have entered the mainstream.

“I was kind of surprised that there’s such a widespread distrust; like we had this question about whether businesses are going to use AI responsibly, and it was pretty overwhelmingly cynical,” Giansiracusa said. “So that kind of struck me, that…distrust of how AI will be used has kind of permeated beyond the tech critics to the larger mainstream culture.”

AI need not apply: Americans have more faith in AI’s ability to do things like customize the content they see online (38% said AI could perform that task better than a human), recommend products or services (27%), or help with homework or studying (26%). They doubted AI’s capabilities most when it comes to dispensing financial or medical advice, driving passengers in a car, or recommending which candidates a company should hire.

Giansiracusa said the ability for the average person to tinker with programs like ChatGPT means they could have a better understanding of the pitfalls than they might for more arcane technologies.

“The overall narrative has emerged in the public awareness that these things are really amazing a lot of the time, and they’re amazingly naïve and stupid some of the time,” Giansiracusa said. “So when you think about businesses using it, it’s like, ‘Well, if my doctor knew a lot and said really smart things, but every now and then said totally stupid things, that’s kind of scary.’ So I think people need more trust.”

A duty to explain: While there’s a tendency for businesses to see awareness as the solution to trust problems, Giansiracusa said he thinks the technology itself needs to become more transparent and grounded in facts.

“There are market pressures that are emerging that this stuff needs to be more reliable, explainable, trustworthy,” he said. “There’s clearly a lot of interest and money and hype. But I think to get that next level of commitment to get these things in production, it needs to go from being experimental to very reliable.”