What if artificial intelligence could find better answers to climate change than we ever imagined? A new system inspired by evolution may be the key to reshaping global land use with unprecedented precision.
Key Points at a Glance
- Researchers trained evolutionary AI to optimize land use and climate policy
- The system balances carbon storage, economic costs, and human needs
- AI-generated strategies revealed surprising, nuanced solutions
- The model is now part of global decision-making tools like En-Roads
- Technology has also been adapted for pandemic response and other crises
Using nearly two centuries of land use and carbon data, a team at The University of Texas at Austin and Cognizant AI Labs has developed an AI system capable of crafting environmental policies smarter than most humans could imagine. At its heart lies evolutionary AI—a digital system inspired by natural selection that fine-tunes policy combinations across thousands of scenarios to yield optimal environmental and economic outcomes.
What sets this technology apart is its ability to navigate complex trade-offs. It doesn’t simply prioritize carbon storage or economic efficiency—it does both, and more. By analyzing how different land use strategies impact carbon emissions and local economies, the system can uncover unexpected pathways to sustainability. As Risto Miikkulainen, one of the project’s leaders, puts it: “There’s always a cost, but AI can help us pick the smartest battles.”
The AI doesn’t settle for the obvious. While converting land to forests might seem like a universal fix for carbon storage, the system discovered that context matters immensely. Replacing croplands with forests, for instance, often yields better climate results than reforesting rangelands. Latitude also plays a role—what works in one region may not in another. Rather than applying broad solutions, the AI emphasizes strategic, localized change.
To build the system, researchers used two core tools: a new dataset tracking global land use back 175 years, and a model that correlates those patterns with carbon flux. This enabled them to create both a prediction model, which forecasts carbon impacts, and a prescription model, which suggests optimal land use strategies for reducing climate change.
The resulting AI recommendations were sometimes surprising even to the scientists. It found better-than-expected results in areas where traditional human policy might hesitate to act. In essence, the system teaches us to optimize the right things in the right places, offering policymakers a lens to act smarter—not just faster.
But the implications go beyond land. The team is expanding its reach through a nonprofit called Climate Interactive, layering this AI onto En-Roads—an interactive climate simulator used by governments, businesses, and educators. Users can now plug in goals, like limiting global warming to 1.5°C, and let the AI recommend policy combinations to get there. This replaces trial-and-error with intelligent strategy.
The success of this tool in land use follows earlier applications of evolutionary AI in public health. During the COVID-19 pandemic, the team used similar methods to recommend lockdown strategies in Iceland, tailored to the country’s unique situation. Their work informed school reopening plans and even helped spark an XPRIZE competition to crowdsource AI solutions to global health crises.
What ties these projects together is a bold idea: that many of the world’s hardest problems—pandemics, climate change, food insecurity—can be tackled with the same powerful engine of intelligent optimization. That engine is now open source, thanks to the team’s Project Resilience platform.
By giving researchers and policymakers access to these tools, Project Resilience democratizes the ability to make better decisions. Whether it’s reducing emissions or preventing the next crisis, AI isn’t just analyzing data—it’s empowering humanity to act more wisely.
For Miikkulainen and his collaborators, it’s a glimpse of a future where AI doesn’t replace humans—it helps us become better stewards of our planet. And that might just be the most hopeful breakthrough of all.
Source: University of Texas at Austin
