A large majority of American voters are skeptical of the argument that the United States should develop increasingly powerful artificial intelligence to compete with China, unfettered by domestic regulations, according to a new poll obtained exclusively by TIME.
The results show that American voters disagree with a common tech industry narrative in which CEOs and lobbyists have repeatedly argued that the U.S. must tread carefully in regulating AI to avoid handing the advantage to its geopolitical rival. And they reveal a surprising level of bipartisan consensus on AI policy, with both Republicans and Democrats supporting the administration’s need to set limits on AI development for the sake of safety and national security.
According to the poll, 75 percent of Democrats and 75 percent of Republicans believe that a “carefully controlled approach” to AI – preventing the release of tools that terrorists and foreign adversaries could use against the U.S. – is better than “moving as quickly as possible on AI to be the first country to get extremely powerful AI.” A majority of voters support stricter security practices at AI companies and are concerned about the risk of China stealing their most powerful models, the poll shows.
The poll was conducted in late June by the AI Policy Institute (AIPI), a US nonprofit that advocates for a “more cautious path” in the development of artificial intelligence. The results show that 50% of voters believe the US should use its lead in the AI race to prevent any country from building a powerful AI system by enforcing “safety restrictions and aggressive testing requirements.” In comparison, only 23% believe the US should try to build powerful AI as quickly as possible to overtake China and gain a decisive advantage over Beijing.
The poll also suggests that voters are fundamentally skeptical of “open source” AI, the idea that technology companies should be allowed to publish the source code of their powerful AI models. Some tech experts argue that open source AI encourages innovation and weakens the monopoly power of the largest technology companies. Others say it’s a recipe for danger as AI systems become more powerful and unpredictable.
“In my opinion, it is not possible from the polls to stop AI development,” says Daniel Colson, executive director of AIPI. “But giving industry free rein is also seen as risky. So there is a desire for a third way. And when we present that in the polls – that third way, a tempered AI development with safeguards – then that is the one that people overwhelmingly want.”
The poll also shows that 63 percent of American voters believe that exporting powerful AI models to potential U.S. adversaries such as China should be illegal. Among Republicans, the figure is 73 percent and among Democrats, 59 percent. Only 14 percent of voters disagree.
The poll surveyed a sample of 1,040 Americans representative of education level, gender, race, and the political parties for which respondents voted in the 2020 presidential election. The margin of error reported for the results is 3.4% in both directions.
So far, there is no comprehensive regulation of AI in the U.S. The White House is encouraging various government agencies to regulate the technology themselves within their existing jurisdictions. But that strategy appears to be under threat from a recent Supreme Court ruling that limits the ability of federal agencies to apply blanket rules set by Congress to specific or new circumstances.
“Congress has been so slow to respond that there is a lot of interest in delegating authority to existing or new agencies to increase government responsiveness” when it comes to AI policy, Colson says. “This (ruling) definitely makes that more difficult.”
While federal AI legislation seems unlikely anytime soon, let alone before the 2024 election, recent polls from AIPI and other institutes suggest that voters are not as polarized on AI as they are on other national issues. Previous AIPI polls found that 75% of Democrats and 80% of Republicans believe U.S. AI policy should aim to prevent AI from quickly achieving superhuman capabilities. The polls also showed that 83% of Americans believe AI could inadvertently cause a catastrophic event, and that 82% favor slowing AI development to account for that risk, compared to just 8% who favor accelerating development.
