A new study suggests that an AI-powered browser extension can soften hostility between supporters of opposing political parties during major elections. Researchers in the United States tested the tool during the 2024 US presidential race and found that even small shifts in what users see online can influence political feelings.
Researchers track algorithm impact
The extension allows independent scientists to examine how social media ranking systems shape public views. It works without help from platforms, offering a rare look at hidden algorithms that guide online discussion. The tool scans posts on X for anti-democratic themes or extreme partisan messages before moving them lower in a user’s feed.
Testing during a heated campaign
A team from Stanford University, the University of Washington, and Northeastern University trialled the tool with more than 1,200 volunteers. Participants agreed to have their X feeds altered for ten days shortly before the election. Some received more hostile posts, while others saw them pushed far down their timelines.
Small shift, big meaning
Those shown reduced hostility reported warmer feelings toward the opposing party, improving by an average of two points on a 100-point scale. Researchers noted that this shift equals about three years of change in national political attitudes. The effect appeared across both liberal and conservative users.
Potential for wider use
Experts said the findings show how algorithm changes could ease online division. They also warned that the study only covered browser users and offered no insight into long-term results. Even so, the researchers believe the tool shows how reranking content could support healthier democratic debate.
