NURTURING A RESILIENT EARTH
Accelerating climate solutions through AI
By: Liz Do & Alison Rutka
24 Oct, 2023
The effects of climate change can be felt globally at an alarming rate. CIFAR is at the forefront of addressing it. This special series, Nurturing a Resilient Earth, takes a closer look at how CIFAR’s community — across borders, roles and disciplines — is making a meaningful impact to bring about powerful change.
Find out how Canada CIFAR AI Chair David Rolnick is deploying AI to accelerate discoveries that could help fight climate change.
Want to learn more about their research?
Watch: CIFAR Talks: Nurturing a Resilient Earth featuring researchers from the series.
The growing ubiquity and sophistication of AI technologies has led to deep debates on ethics and their world-changing effects on society. Among the areas in which AI adoption could play a crucially positive role is the fight against climate change.
David Rolnick, a Canada CIFAR AI Chair at Mila, is the co-founder and chair of Climate Change AI, a global non-profit focused on the intersections of climate change and machine learning. A recent report produced by the group in collaboration with the Global Partnership on AI makes recommendations for governments about the use of AI for climate action, targeting four key areas in which machine learning can accelerate progress: distilling raw data into actionable information; improving predictions; optimizing complex systems; and accelerating scientific modelling and discovery.
The report helped to inform a recent planning symposium hosted by CIFAR’s Pan-Canadian AI Strategy, which brought together Canadian stakeholders from across academic, government, private and philanthropic sectors to inform the next steps for the strategy in advancing the use of AI for sustainable energy and the environment.
According to Rolnick, the use of AI to advance climate goals could take on numerous applications. From monitoring biodiversity to designing new materials for renewable energy projects, to assessing crop yield from satellite imagery, he and his team are using AI in new and inventive ways to complement existing research methodologies.
“There is never going to be a world in which we don't need physical experimentation or don't fundamentally rely on knowledge of chemistry and material science,” says Rolnick. “But sometimes AI can speed up the process of intuition or gather data at scale to enhance the research process. AI is only one piece of the puzzle.”
Rolnick is also working to advance our ability to model the climate. Current climate models use known physical laws to simulate climate predictions with a high degree of accuracy. However, given the vast amounts of information being processed, this can move at a snail’s pace. “Sometimes these models take months to run even on supercomputers,” he notes.
Rolnick’s AI algorithms are speeding up climate models used by the Canadian government. Working closely with Environment and Climate Change Canada, Rolnick uses machine learning to take pieces of existing climate simulations and approximate them less accurately, but much faster.
Recently, Rolnick’s team applied AI to a model for radiative transfer, a way of measuring the transfer of electromagnetic radiation in the atmosphere. They were able to produce an approximate simulation of radiative transfer in a much more condensed time frame.
“AI is essentially a way to cut a corner in the simulation,” he says. “Our tools have worked so well that they're actually in the process of being integrated into Environment and Climate Change Canada’s national climate models.”
Despite these advances, Rolnick admits there are challenges in adopting powerful technologies like AI for global application to solve the world’s problems. In particular, he acknowledges the many bottlenecks that exist in leveraging machine learning — everything from digital infrastructure to deployment capacity, to building and funding new technologies.
International collaboration is one such bottleneck in which AI has the potential to unlock immense efficiencies, through the sharing of data at a global, collaborative level. As a Canada CIFAR AI Chair, he’s seen the power and potential impact of knowledge and data-sharing.
“Being a Canada CIFAR AI Chair has greatly helped my work in machine learning for climate action, by providing flexible funding that can be used for projects in traditionally underfunded application areas such as ecology,” says Rolnick. “CIFAR has also enabled a vibrant Canadian AI ecosystem with close ties to the private and public sectors.”
Rolnick is motivated to continue contributing his AI expertise in the fight against climate change.
“I want to use my knowledge and skills in a way that is meaningful in impacting arguably the greatest problem that humanity faces,” he says. “That’s what led me to this intersection between different fields.”
How does CIFAR help you do impactful work to address climate change?
“Being a Canada CIFAR AI Chair has greatly helped my work in machine learning for climate action, by providing flexible funding that can be used for projects in traditionally underfunded application areas such as ecology. CIFAR has also enabled a vibrant Canadian AI ecosystem with close ties to the private and public sectors.”