Americans Divided on Social Media’s Use of AI to Fight Misinformation

About seven-in-ten Americans use social media platforms to connect with others, share aspects of their lives, and consume information. But the content they encounter is increasingly shaped not just by their own choices, but by sophisticated algorithms and artificial intelligence technologies deployed by these platforms.

A new Pew Research Center survey reveals Americans are deeply split over social media companies’ use of algorithms to identify false information—with 38% viewing this as a positive development for society, 31% considering it negative, and a similar share uncertain about its impact.

These algorithms serve multiple functions on social media platforms: determining what content users see in their feeds, targeting advertisements, making content recommendations, and helping identify and remove problematic content including hate speech and misinformation.

The technology is clearly visible to most users—74% of social media users report having seen content flagged or labeled as false on these platforms. Three-quarters of Americans say they’ve heard at least something about these computer programs, though only 24% report having heard a lot about them.

Sharp partisan divides characterize attitudes toward these tools. Republicans are significantly more likely than Democrats to view algorithmic content moderation as harmful. For example, 84% of Republicans believe political viewpoints are being censored due to these algorithms, compared to 56% of Democrats. Similarly, 81% of Republicans versus 55% of Democrats believe news and information are being wrongly removed from platforms.

“Even with some across-the-aisle agreement, there are stark partisan differences on the potential impacts we explored,” noted the researchers. “Republicans are far more likely than Democrats to say the widespread use of these algorithms is a bad idea for society.”

The research found a widespread belief that algorithms are negatively impacting the online information environment. Seven-in-ten adults say political viewpoints are being censored, and 69% believe legitimate news and information are being wrongly removed. Meanwhile, only about four-in-ten believe these technologies are enabling more meaningful online conversations or making it easier to find trustworthy information.

When it comes to how these systems should operate, Americans express clear preferences. A strong majority (69%) believe accuracy should be prioritized over speed in content moderation decisions, even if it means some false information remains online longer. Only 28% prioritize quick decisions, even at the risk of mistakenly removing accurate information.

Most Americans (71%) also believe decisions about what constitutes false information should involve human judgment—either exclusively or in combination with algorithms. Only 6% think these decisions should be made primarily by computer programs alone.

The survey reveals limited confidence in social media companies’ use of these tools, with 72% of Americans expressing little or no confidence that platforms will use algorithms appropriately to identify false information. A significant majority (53%) worry government regulators will not go far enough in regulating these technologies, while 44% fear excessive regulation.

Looking beyond social media to other algorithmic decision-making contexts, Americans express strong opposition to fully automated systems making final decisions in several high-stakes situations: 70% oppose algorithms having the final say on medical treatments, 64% on parole decisions, 60% on job applicant screenings, and 56% on mortgage approvals.

The findings highlight the complex challenges facing both technology companies and policymakers as they navigate public concerns about algorithmic systems that increasingly shape our information environment and many aspects of daily life.