An artificial intelligence model in a recent European study predicts that 1.1 billion people could expose to food crises by 2100 due to climate change’s direct and indirect effects.
The climate analysis, conducted by the European Union’s Joint Research Centre (JRC) and published in Scientific Reports, shows that conflicts and inequalities caused by climate change could more than triple the likelihood of food crises by the end of the century, compared to scenarios involving sustainability and mitigation approaches.
The study, seen by Food & Climate, reveals stark differences in the future of food crises depending on socioeconomic trajectories.
From climate alone shows large gaps across socio-economic pathways: conflict and inequality could more than triple end-century exposure to food crises compared to sustainability and mitigation scenarios.
The findings show that by 2100, conflict- and inequality-driven scenarios could expose over 1.1 billion people — including more than 600 million children, mostly in Africa and Asia — to severe food crises.
In contrast, sustainable pathways could more than halve the exposure, highlighting how policy choices determine whether hundreds of millions face crises or far fewer are affected.
Predicted food crises based on temperature and rainfall data
The new AI model can predict food crises based solely on climate data.
The projections are based on a powerful AI model – developed by the author – capable of identifying the onset of new, severe food crises based only on temperature and precipitation data.
The model was trained and tested on a large dataset of historical acute food insecurity data for the period 2010-2022, obtained from the Famine Early Warning Systems Network (FEWS NET).
The high accuracy of the model corroborates the idea that climate plays a major, if the not the main role in food-security, not only through obvious direct effects – such as droughts reducing crop yields – but also through indirect, more convoluted pathways, by modulating, for instance, drivers of conflict, migration, and disease.
By relying solely on widely available climate data, it provides a scalable tool for exploring long-term food-crisis scenarios, unlike traditional models requiring extensive socio-economic inputs.
FEWS NET data show that exposure to severe food insecurity nearly tripled from just over 50 million in 2011 to almost 150 million by 2020.
When projected onto future climate simulations and combined with demographic and poverty projections consistent with the IPCC Shared Socioeconomic Pathways (SSPs), the AI model reveals sharply diverging food-crisis risks by century’s end.
Adopting a Sustainability Approach
Under a “Sustainability – taking the green road” pathway (SSP1), average yearly exposure drops by about 75% to 42 million.
In contrast, conflict and inequality futures raise yearly exposure respectively to 280 million (SSP3) and 229 million (SSP4).
Compared with SSP1, these scenarios put five to seven times more people at risk, and even a middle path (SSP2, “business as usual”) more than doubles exposure.
Beyond annual risk, the cumulative exposure is staggering: more than 1.16 billion people experience at least one famine crisis this century under the inequality pathway (SSP4), whereas the sustainability pathway could spare about 780 million by 2099.
The burden falls heavily on the young – most exposed individuals are children under five at their first crisis (about 630 million under SSP3–SSP4), and hundreds of millions of newborns face a crisis within their first year (230–270 million).
Future food-crisis risk is highly uneven across regions. Although the model detects emerging risks beyond today’s monitored areas, most projected crises occur in already vulnerable regions, especially Africa and Asia.
Africa faces the widest geographic spread of crises, while Asia’s higher population density leads to comparable numbers of people exposed. Under conflict and inequality pathways (SSP3–SSP4), exposure rises steadily on both continents. Under more sustainable futures, however, trajectories diverge: Africa sees a marked decline in exposure after mid-century, while Asia’s exposure remains largely stable. This suggests that even optimistic pathways may deliver limited improvements for Asia’s food security compared with Africa.
Climate change alone does not cause humanitarian crises: social and political conditions act as a multiplier. Societies with high inequality, limited development, or those suffering from ongoing conflict are far more vulnerable, even under the same climatic hazards.