“Humanity needs to scale up the good AI solutions for our planet”

The rapid development of AI is taking place in an emergent crisis landscape, ranging from record-breaking temperatures and loss of nature to economies and societies in fragile states.

In January, a global trans-disciplinary team of researchers met in Stockholm to explore how novel AI methods can help us understand and tackle these connected sustainability risks.

“AI is a huge experiment. It can do amazing things, but history shows that new technology could also cause damage. We need to scale up the good solutions for our planet, this will not happen by itself”, says Victor Galaz, who co-led the workshop “Seeing Earth through AI”, and has led the AI research at Stockholm Resilience Centre at Stockholm University since the start in 2020.

Victor has earlier authored a science report on AI and misinformation, and has just published a new book. In Dark Machines (Routledge), he explains and explores why AI and automation may amplify risks for people and the planet.

“We face a profound challenge. We can either allow AI to accelerate the loss of resilience of people and our planet. Or we can decide to act forcefully in ways that redirects its capabilities towards resilient, sustainable, and just futures,” he says.

In Seeing Earth through AI, the research team will now explore this rapidly developing AI field with a focus on the potential of AI methods to help inter- and transdisciplinary researchers develop science insights with pertinence for sustainable development.

“AI can help fight climate change and build a more sustainable, low-carbon world. It´s crucial that as many people as possible now engage to support AI development on sustainability,” says Drew Purves, sustainability and biodiversity co-lead at Google DeepMind.

The Seeing Earth through AI team will develop an open access report to assist the application of AI for sustainability science – that is, how it can be used for augmenting monitoring, predictive modeling and evidence-based decision-making.

“Modern AI methods can make substantial contributions to deepen our understanding of the global systemic risks and transformative opportunities in the intertwined human-Earth system in the Anthropocene, including improving simulation modeling and reducing uncertainties on tipping point risks in the climate system,” says Jonathan Donges, researcher at PIK, and one of the co-leads of the project.

“These new AI tools need to be applied in a responsible, safe and scientifically sound way to leverage their full potential for accelerating research progress and driving social-ecological transformations, avoiding pitfalls and adverse effects,” Jonathan Donges adds.

In a first workshop held in Stockholm at the start of 2025, the researchers shared insights in this emerging field. The team raised opportunities and challenges using AI for scientific analysis, including ethical considerations and use of energy.

Solutions to scale include improved understanding of climate and social tipping points, how communities are impacted by extreme weather events, monitor loss and growth of nature, speed up work in global science panels like IPCC and support policy processes like the biodiversity COP.

Victor Galaz and Drew Purves both underline that societies now need to mobilize resources quickly to make sure that AI development and usage contributes to critical questions in the sustainability sciences.

“Science is not fully equipped to address and understand the implications and possible applications of AI. We need to think big and collaborate more closely,” says Victor Galaz.

“Sustainability solutions need to address complex interactions among climate, nature, and people, from local to global scales. AI has the potential to be revolutionary here, helping to combine complex data, create new predictive models, and empower decision makers in all sectors,” Victor Galaz adds.

“We can be inspired by how the physics community came together to design and build the Large Hadron Collider, enabling them to address some of the most fundamental questions around particle physics. AI has the potential to enable the sustainability community to join up, share knowledge and data, and truly work together on addressing our most pressing environmental challenges," Drew Purves says.