A new study published on 31 October in The Journal for Artificial Societies and Social Stimulation, a peer-reviewed academic journal, may have shed some light on the root cause of religious violence and how to potentially manage it (1). This is the first time theories on the human mind have been programmed into a computer. The aim of this research was to study the escalation of violence over time and subsequent détente, and how this could be controlled.
The international team of researchers from Oxford in the UK, Boston University and Tufts University in the US, and the University of Agder, Norway used something called multi-agent artificial intelligence (AI) ― different to machine learning ― to create a psychologically realistic model of human society based on cognitive psychology theories. They looked specifically at the conditions that allowed two cases of xenophobic social anxiety to escalate into extreme physical violence: the Northern Ireland Troubles and the 2002 Gujarat riots of India.
The conflict in Northern Ireland, involving the British army and various Republican and Loyalist paramilitary groups, lasted 30 years and was one of the most violent periods in Irish history, claiming the lives of 3,500 and injuring a further 47,000 people. While significantly shorter, the Indian conflict was a devastating 3 days of violence between Hindu and Muslim communities in Gujarat, India’s Westernmost state, that led to the deaths of 2,000 people.
The agent-based model was used to better understand the emergence and escalation of xenophobic anxiety between individuals from these two different religious groups within an artificial society. First, the team trained the AI agents to mimic how human beings would naturally think and process information. Then they created a simulated environment of the real world and filled it with hundreds of these human model agents to represent everyday society and how people of different religions interact.
Mutually escalating xenophobic anxiety ― an increase of the average level of anxiety of individuals in both groups over time ― can reach “a boiling point” leading to violence. Interestingly, only 20% of the simulated scenarios escalated to violence, triggered either by people either outside or within the group going against the core beliefs and identity of the group.
The model does have some limitations related to a number of assumptions made during the modelling process but will no-doubt serve as a platform for future work in this area. The team has already received further funding for projects, based at the Center for Modeling Social Systems in Kristiansand, Norway, related to demographic shifts caused by immigration and the integration and resettlement of refugees in Europe.
Importantly, the “causal architecture” of this type of model allows macro-level phenomena to be predicted from micro-level behaviours and interactions. In other words, we may be able to understand how large-scale religious violence is triggered at an individual level, which could potentially be used to prevent future conflicts. In the wrong hands, this information could, unfortunately, be used to incite violence; however, used appropriately, models like this could provide a positive tool for supporting stable and peaceful societies.
(1) Shultsa, L. et al. A Generative Model of the Mutual Escalation of Anxiety Between Religious Groups. The Journal for Artificial Societies and Social Stimulation (2018). DOI: 10.18564/jasss.3840