Most American voters do not want companies to build superintelligent AI, according to a new survey reported by Tech.co. Public affairs firm Bryson Gillette polled 1,000 registered voters at the end of September and found 58% opposed to building such systems, with the midterm elections days away.
The survey defined superintelligent AI as systems smarter than humans at nearly everything, including scientific research, strategy and persuasion. 61% of respondents said humans would keep little or no control over it once built.
Fear of losing control
Concern runs deep. 83% of voters said they worry about losing control of powerful AI, and 73% worry it could threaten human existence. Lead researcher Lindsey Mears told Tech.co that people see the technology "advancing faster than the accountability around it."
Tech.co notes that the Trump administration has largely stayed away from federal AI regulation. A September summit with industry leaders produced no new federal law.
A campaign issue
Mears said AI has moved from a technology story to a campaign topic, with data centers, electricity costs and control of these systems debated on the trail. In her view, whichever party holds Congress in 2027 will set the terms for AI policy, and she expects more political scrutiny after the vote because skepticism crosses party lines.
The trust gap
Two-thirds of voters say tech and AI companies have real influence in Washington, but only 38% trust them. Jobs feed that distrust: 58% are concerned about AI-driven job losses in general and 44% about their own job.
Mears advises companies to be specific about which AI tools they use and where a person makes the final call, to pair rollouts with training, and to keep productivity claims modest.