The 21st century has provided ample evidence of the far-ranging effects of armed conflict. Beyond the obvious toll of immediate death and destruction, the interconnectedness of today’s world means there can be significant political, economic and social repercussions far from the fighting. As such, predicting conflicts — with the aim of preventing them — is a stated objective of organisations such as the United Nations and World Bank.

Future Horizons:

×××

5-yearhorizon

Machine-learning models improve conflict risk-detection

The continued development of machine-learning models and other data-driven systems helps to improve detection of increased risk of conflict. The models are validated against historical and current experience to prove that they produce accurate predictions of the effects of conflicts and any interventions made to prevent them.

10-yearhorizon

“Explicable” machine learning facilitates better understanding of conflict prediction

The development of explicable machine-learning systems allows their predictions to be better understood. Integration with theoretical models that capture top-down forces (such as communications from political elites) helps to refine predictions and shape interventions. The first real-world interventions (partially) based on such predictions occur.

25-yearhorizon

Conflict-prediction models adopted by global organisations

Endorsement and adoption of these approaches by global multilateral organisations, and development of accompanying ethical and practical frameworks to ensure that predictions instil confidence rather than hastening conflict and that interventions are timed and designed to respect the autonomy of those affected.

This is not simple. It is easy to identify places whose social order is fragile or which are geopolitically exposed, but much more difficult to determine if and when this will tip into conflict, especially in countries with a long previous history of peace. Attempts are now being made to do this using machine learning, which recognises patterns in data to predict likely outcomes. However, this requires massive amounts of data. One such effort, for example, uses systems trained on millions of news articles dating back to the 1980s.⁴

Purely data-driven approaches, however, may or may not map onto prevailing theories of conflict escalation or outbreak. Nor do purely data-driven predictions offer any guidance as to the potential form of any intervention. Finally, there is also the inherent problem that any publicly disclosed prediction can itself influence the course of events by affecting public sentiment or government decision-making, or by making transparent the thresholds at which intervention becomes likely. Additionally, the use of large datasets in conflict areas raises potentially thorny questions related to surveillance and the potential misuse of data by combatants.

Predicting the onset of armed conflict - Anticipation Scores

The Anticipation Potential of a research field is determined by the capacity for impactful action in the present, considering possible future transformative breakthroughs in a field over a 25-year outlook. A field with a high Anticipation Potential, therefore, combines the potential range of future transformative possibilities engendered by a research area with a wide field of opportunities for action in the present. We asked researchers in the field to anticipate:

The uncertainty related to future science breakthroughs in the field
The transformative effect anticipated breakthroughs may have on research and society
The scope for action in the present in relation to anticipated breakthroughs.

This chart represents a summary of their responses to each of these elements, which when combined, provide the Anticipation Potential for the topic. See methodology for more information.