Machine learning reveals: global migration far larger than previously thought

A new deep learning model developed by researchers at the London School of Economics and Political Science and the University of Hong Kong shows that global migration has grown from 13 million to nearly 35 million people annually. The model highlights that changes occurring in the global South are far more extensive than Western perspectives suggest.

Researchers at the London School of Economics and Political Science (LSE) and the University of Hong Kong have created a new deep learning model that reveals the true scale of global migration for the first time. According to the model, global migration has grown from approximately 13 million people since the turn of the millennium to nearly 35 million annually, a growth whose magnitude has largely gone unnoticed.

According to the researchers, what stands out most is that the most significant change is occurring in the global South. Regions often sidelined in Western media perspectives have actually experienced the most dramatic migration developments. The use of machine learning enabled analysis of large volumes of data that were previously difficult for researchers to access or were in raw, unprocessed form.

Why earlier data fell short

Previous migration statistics have often relied on data collected by national authorities, which fail to capture numerous movements, particularly undocumented or unregistered migration. The new model is able to identify patterns even in regions where official registration is lacking, bringing to light millions of people who were left in statistical blind spots.

The application of deep learning methods in migration research represents a significant breakthrough in the social sciences. Similar models are increasingly being used to make forecasts related to climate migration and to plan humanitarian aid, where accurate data is of critical importance.

What the results mean

The study emphasizes the need for a broader perspective on global migration than previously adopted. When policymakers rely on outdated or incomplete data, many critical questions, from healthcare to educational opportunities, remain insufficiently addressed. According to the researchers, more accurate data is essential for both international organisations and national governments that shape migration policy for the future.