Common global narratives about political polarization are skewed because of an over-reliance on literature around the US, leaving governments unprepared on how to tackle the problem at home.
In new research, led by King’s College London and published in Nature Communications , experts across the UK, Europe, Asia, Australia and the US point to how common narratives about polarisation are overgeneralised and often based on evidence from the US alone.
Academics warn that this risks applying US solutions to non-US problems, with local policymakers having unsuitable evidence to tackle the growing polarisation challenges of social media and AI, with dangerous ramifications for democracy.
In the context of AI and social media, these wrong assumptions can lead policymakers to make decisions that fail to address the true drivers of polarisation. Dr Max Falkenberg, AI+ Senior Fellow at King’s College London and lead author of the paper, said: “Social science has an America problem. The first Trump Presidency and the role of social media in that election caused an explosion in polarisation studies in the US that the rest of the world is yet to catch up with. But norms established in the US such as political antagonism being based on a one-dimensional left-right axis, don’t necessarily fit with what’s happening in other parts of the world.
“Ensuring a global literature on political polarisation must be a fundamental goal of the academic community. There are legitimate concerns around AI being used as a tool to further polarise our societies, but understanding the polarizing role of AI will only be possible if the evidence is not solely US-focused. If we fail to understand the nature of political disagreement, how can we hope to be better?”
Acknowledging how US political actors like Trump still have an influence in global domestic politics, Dr Falkenberg and other critics point to how resolutions such as the Good Friday Agreement in Northern Ireland and separatist disarmament in Spain complicate folk narratives of ever-increasing global polarisation.
They note the imbalance of academic journals either being US-based or entirely US-focused has also impacted the very definition of political polarisation away from being applied in a European and global context, with many US-focused papers not even mentioning their subject explicitly.
Selecting ‘polarisation’ as their word of the year in 2024, Merriam-Webster defined it as “division into two sharply distinct opposites; especially: a state in which the opinions, beliefs, or interests of a group or society no longer range along a continuum but become concentrated at opposing extremes”.
While the definition may fit into two-party political systems in the US, it is unclear whether the same conceptions of political animosity and grouping of parties happen in the multi-party democracies and political power sharing seen in continental Europe.
While current narratives suggest social media algorithms and misinformation to be a major cause of political polarisation, this is heavily disputed and unlikely to be the case in a number of countries. For example, a study that deactivated Facebook in the US reduced polarisation, but when replicated in Bosnia-Herzegovina researchers found the opposite occurred.
By supplying an alternative to individuals’ lack of offline exposure to other ethnicities Facebook helped promote political cohesion. This suggests that any move to regulate social media, and by extension the use of AI in recommendation algorithms, must be country dependent as opposed to a singular universal rule informed by a US context.
Professor Andrea Baronchelli, Professor of Complexity Science at City St George's University of London and the paper's senior author, said " Polarisation is not a single phenomenon with a universal set of causes or solutions. Political institutions, social identities and media environments differ across countries, and the same intervention can have very different, even opposite, effects in different contexts. If we want effective policy, we need to understand those differences rather than assume that one model of polarisation applies everywhere.”
As a way to counteract the over-reliance on the US, the researchers advocate the promotion and funding of work outside the US and comparative research that captures the commonalities and uniqueness in the polarisation of different countries. They also suggest that academia push back against social media companies' efforts to restrict data access as a necessity to establish links between polarisation and digital media use across the globe.
Nature Communications
2-Sep-2026