A significant majority of American voters prefer a cautious approach to AI development over a rapid race to outpace China, according to a new poll shared with TIME. The findings suggest a bipartisan consensus on AI policy, with both Republicans and Democrats favoring government-imposed limits on AI development for safety and national security.
The poll, conducted in late June by the AI Policy Institute (AIPI), found that 75% of both Democrats and Republicans support a “careful controlled approach” to AI. They prefer preventing the release of tools that could be exploited by terrorists or foreign adversaries over moving quickly to build the most powerful AI systems. Additionally, most voters favor stringent security practices at AI companies to mitigate the risk of China stealing advanced AI models.
Daniel Colson, Executive Director of AIPI, highlighted the public’s desire for a balanced approach: “Stopping AI development is not seen as an option, but giving industry free rein is also risky. People overwhelmingly want a third path: mitigated AI development with guardrails.”
The survey also showed that 63% of voters believe exporting powerful AI models to adversaries like China should be illegal, with higher support among Republicans (73%) than Democrats (59%). Only 14% opposed this idea. The representative sample of 1,040 Americans revealed a margin of error of 3.4%.
These findings emerge amid a lack of comprehensive AI regulation in the U.S. The White House has encouraged agencies to self-regulate within their existing remits, but a recent Supreme Court ruling has limited federal agencies’ ability to apply broad regulations to new technologies, complicating the regulatory landscape.
Despite federal AI legislation appearing unlikely before the 2024 election, the poll indicates voters are not as polarized on AI as on other national issues. Previous AIPI polling found that 75% of Democrats and 80% of Republicans believe U.S. AI policy should prevent AI from quickly reaching superhuman capabilities. Furthermore, 83% of Americans fear AI could accidentally cause a catastrophic event, with 82% supporting a slower development pace to address these risks.
