Digital “psychological twins” promise a new way to understand social trends without intruding into people’s lives. The creator of the tool, developed from experiences drawn from conflicts in Northern Ireland, emphasizes: it is not about predicting individual crimes, but about understanding the dynamics of societies. The goal is to test policies or disinformation narratives in a safe synthetic environment, not on real people.
Psychological twin: how it works and why it isn’t just about language
At the core of the technology is an algorithm that creates a psychologically realistic “digital twin” of a person with high information compression, allowing it to scale to millions of individuals. Unlike large language models, which return an average, this approach looks for variability in the population—including “pockets” of extremism or polarization. It draws on diverse data sources in more than 200 languages: from social networks and media to key in‑person interviews directly in the field, where people describe their motivations more openly. The result is “agents”—for example, in the forthcoming digital twin of Slovakia there will be more than five million of them—on which responses to policies or disinformation influences can be simulated without risk to real society.
Accuracy, safety, and a surprising finding about people
The team states that it integrates into the models only those psychological causal effects for which it achieves a 95% confidence interval, and that academic institutions have audited the work. They store sensitive data under maximally secure conditions, and in high‑risk cases outside the online environment. The first case also comes from the field: contact with a Loyalist paramilitant in Belfast showed an interest in disarmament and a shift to political tools, which confirmed the value of modeling the path to peace. Most surprising, however, was the realization that conflict is not the “default state” of societies—most of the time cooperation prevails, and violence occurs when ordinary people find themselves in extraordinary circumstances; for this reason the authors also warn against possible misuse and carefully choose clients (from international organizations to governments), while for Slovakia they see scenarios of improvement as well as the risk of a temporary worsening of polarization.
