One of the nice things about covering the frontier of technology — large language models, quantum, virtual worlds — is that they’re decidedly less partisan than most policy issues. Sure, there might be disagreements, but these topics largely lack the clear battle lines, deep trenches or scorched-earth center of subjects like health care or climate change. That might be changing. A new round of polling about voters’ opinions of artificial intelligence — shared exclusively with Digital Future Daily by the AI Policy Institute — reveals early partisan and demographic splits taking shape, all with the first election to take place in an AI-saturated media environment looming on the horizon. The poll conducted by AIPI, a recently formed nonprofit that describes itself as “dedicated to seeking political solutions to potential catastrophic risks from emerging AI technology,” surveyed just over 1,000 voters and found some stark topline results: - 76 percent of voters prefer candidates who support regulating AI
- 55 percent want that approach to be bipartisan
- 61 percent either “strongly” or “somewhat” support the Senate’s current slate of proposals for AI legislation.
But dig deeper into the poll’s crosstabs and you see something all too familiar to American politics: Voters say they want bipartisan solutions, but have serious disagreements about what should be done and why. A few of the key takeaways: Political concern over AI tracks with education. When asked whether “a politician’s positions on artificial intelligence [will] be something you consider when voting in the 2024 Election,” 58 percent of college-educated voters said “yes” while 57 percent without a college degree said “no.” Likewise, while 85 percent of college-educated voters responded affirmatively that “As artificial intelligence develops, AI will become a more important political issue for me,” only 66 percent of non-college voters did. Similar splits emerge when it comes to specific proposals like “legislation that would establish a liability regime for consumer fraud committed using AI,” although both college and non-college graduates broadly agree that the government should be “doing more to regulate AI” — and more specifically that private companies should not be trusted to self-regulate. Overall, in keeping with the increasing reality of education as a proxy for partisanship, this strongly implies that Democrats are broadly more likely to respond supportively to proposals for AI regulation. Despite that… Democrats and younger voters are techno-optimists, more likely to say in this poll that AI will make the world better and specifically improve their lives. When asked whether advances in AI will be good or bad for “people like me,” 60 percent of respondents aged 18-44 and 66 percent of Democrats responded “good.” Asked the same question about “society generally,” Democrats responded in the affirmative by 64 percent while the 18-44 demographic was exactly split. (Despite their ambivalence on the latter question, 59 percent of 18-to-44-year-olds said AI would benefit “me personally.”) Young voters seem even less eager to regulate AI aggressively. Asked whether AI companies should self-regulate, 45 percent of the 18-44 demographic replied “yes,” 11 points higher than the age bracket above them. The youngest age bracket also reported significantly less support for “legislation that would establish a liability regime for consumer fraud committed using AI” and “legislation that creates protections for people who may face discrimination from AI models” than older respondents. That could imply, given younger voters’ normally heavy skew toward Democratic beliefs, that they simply aren’t as afraid of the technology as their older political peers. The partisan divide is faint, but it’s there. As one might imagine from the party of the free market (well, mostly, until very recently), Republicans are generally slightly less favorable toward regulating AI than their Democratic counterparts. But they also take an overall more skeptical view of the technology and its effect on society itself, with 58 percent of Republicans saying it would be bad for “people like me,” 63 percent saying it would be bad for “working class people” (compared with Democrats’ more optimistic 46 percent), and 66 percent saying it would be bad for “middle class people.” (Independents hew largely toward the middle on these issues, with the notable exception that they nearly match Republicans when it comes to their view of AI and the working class, with 59 percent saying it will be bad.) And what about, you know, the actual politicians? Asked who voters “trust most” on AI between President Joe Biden and former President Donald Trump, respondents split predictably among partisan lines with 25 percent of Democrats and 34 percent of Republicans still saying they’re “not sure.” Independents split almost evenly with 22 percent for Biden and 24 percent for Trump, with 49 percent still undecided. AI isn’t likely to turn into a lightning-rod culture-war issue anytime soon. But if it does take on a greater partisan valence, as AIPI’s polling suggests is happening, it’ll inevitably be more difficult to have measured debates, or pass the bipartisan laws being floated in the Senate right now, or get Democratic Senate Majority Leader Chuck Schumer and Elon Musk to agree on something. In that world, AI might seem less like the next chapter in a feel-good story about Silicon Valley dynamism and more like another partisan football.
|