AI and tech

Human enterprise is fundamentally reshaping all life in the biosphere, and the function of the Earth system. AI is as part of other technological innovations, a technology that is likely to change these dynamics, at times in very unexpected ways. A key question is: will it accelerate the “Great Acceleration” or support a just transition? That is still very much an open question, and a choice that needs to be made by society informed by science.

The Centre plays a key role in helping individuals, companies and decision-makers make sense of digital technologies and their interactions with humans embedded in the biosphere. There are substantial governance and regulation challenges which requires us to take a step back from the simplistic “AI for Climate”-narratives. Are the AI solutions from governments and big tech really helping to solve urgent climate and sustainability challenges? Who should AI innovation really be for? And what would it take to advance planetary responsible AI innovations? Centre research will continue to explore these questions in the coming years and focus on better understanding the full picture. This will involve exploring both the very material aspects of a growing AI industry, potential risks to people and the planet, and the immense opportunities that AI methods can bring to the sustainability sciences in close dialogue with society.

Current Centre work on AI and tech

Understanding the influence of AI on our planet implies using AI methods to explore issues of vast importance for our research. The Centre contributes to a research project led by (former Centre PhD-student) Jonas Hentati-Sundberg at the Swedish University of Agricultural Sciences (SLU) that combines video images from CCTV with deep-learning methods to automate the analysis of guillemots’ life and the environment. The team also attempts to use AI to analyse data from a sailing drone which is powered by solar panels and equipped with a range of sensors, to measure for example ocean temperature, plankton blooms, cyanobacteria and turbidity. It sends updates on ecosystem status over satellite every ten minutes. Such approaches can support real-time decision-making in marine ecosystems, and help build a deepened understanding of a part of the Baltic Sea ecosystem over time. AI is useful to study other resilience challenges as well. The Centre has a close collaboration with colleagues at New York University through the project ClimateIQ led by Timon McPhearson. The project explores the potential to augment climate risk models for cities using deep learning methods.

Over the past few years, a group of researchers at the Centre in collaboration with the Beijer Institue of Ecological Economics has also studied the potential to automate climate misinformation on social media platforms through, for example, automated fake social media users known as “social bots”. This work has been led by Stefan Daume and has, as presented in the report for the Nobel Prize Summit 2023, "Truth, Trust and Hope", become very important in times when generative AI (e.g., ChatGPT and Llama) is broadly accessible. The abilities of social bots to generate personalized human-like content, such as text, images, videos and coding, should not be underestimated. It would however, be impossible to study these issues without also using AI to help analyse up to millions of text snippets and other user characteristics. Centre research on online content around extreme weather emergencies, climate change, sustainability and environmental crisis shows that the influence of automation in online climate discussions is vastly more complicated than has been reported in media, with ‘social bots’ only partly driving misinformation and polarisation.

In 2025, the Centre published, "AI for a planet under pressure", a landmark report on how AI can be applied both responsibly and effectively to address complex and interconnected sustainability challenges. The report was the result of a collaboration between the Centre, the Potsdam Institute for Climate Impact research and Google DeepMind.