New study suggests social media algorithms shape opinions more than previously thought

A new study from researchers at the University of Copenhagen and other institutions indicates that even minor changes to social media algorithms can influence political polarisation and the accuracy of users’ factual beliefs, Danish broadcaster DR reports.

The study, published in a university press release, challenges the claim by tech companies that their algorithms merely help users find content they want to see. Instead, the research suggests these systems may actively contribute to polarisation and resistance to factual information.

“Our study shows that the way algorithms function can potentially lead users to become more polarised or less receptive to facts,” said Jason William Burton, assistant professor at the University of Copenhagen’s Department of Psychology and Centre for Social Data Science, who led the study. The research also involved Technische Universität Dresden and the Max Planck Institute for Human Development.

Experiment with US participants

The study involved an online experiment with American participants, split evenly between liberal and conservative political leanings. The first phase included 500 participants, while the second expanded to 1,000.

Researchers tested different algorithm designs, including one modelled after real-world engagement-based systems—such as those used by Facebook and X—which prioritise content generating high interaction (likes, shares, reactions). This algorithm led to increased polarisation and less accurate factual assessments among participants.

Alternative algorithms were also examined: one that promoted “bridge-building” content popular across political divides, and another that targeted posts correcting factual misconceptions. The bridge-building algorithm improved agreement between liberal and conservative participants in some cases, while the accuracy-focused algorithm enhanced factual precision compared to both random and engagement-based rankings.

Call for regulatory action

The findings suggest that alternative algorithm designs could reduce polarisation and misinformation. However, the researchers note that tech companies may resist changes that threaten their business models, making political regulation a potential solution.

“Our experiment shows we should explore more types of algorithms, as they could have positive effects on public discourse,” Burton said.