The technical capabilities demonstrated by researchers suggest new possibilities for how platforms could function. By integrating AI analysis with live feed manipulation, scientists created systems that could identify divisive content and adjust its visibility in real-time. Similar capabilities deployed at scale by platforms themselves could dramatically reshape political discourse for better or worse.
The research team developed AI systems that analyzed X posts during the 2024 presidential election for markers of anti-democratic attitudes and partisan animosity. The system operated in real-time as over 1,000 users scrolled their feeds, constantly evaluating content and adjusting what appeared based on research objectives. This integration of analysis and manipulation happened seamlessly and invisibly from users’ perspectives.
Platforms already possess such capabilities but currently use them primarily for engagement optimization. The same technical infrastructure could be repurposed for other objectives—promoting factual accuracy, reducing polarization, supporting constructive dialogue, or any number of other goals. The research proves that divisiveness can be dialed up or down through algorithmic choices.
This power raises crucial governance questions. Who should decide what objectives algorithms pursue? Currently, platform owners and executives make these choices, optimizing primarily for metrics that generate revenue. But given the massive societal impacts demonstrated by research, should these decisions be subject to democratic input, regulatory oversight, or public accountability?
The AI capabilities will only grow more sophisticated. Future systems might analyze not just text but images, videos, emotional tones, and complex contextual factors. As platform power increases, the urgency of governance questions intensifies. Societies must decide whether to trust corporate decision-makers with such power or to demand transparency, accountability, and potentially democratic control over algorithmic systems that shape public opinion.