CDS’ Grace Lindsay Launches YouTube Channel on “AI for the Planet”
Climate change applications of machine learning often lack the philosophical debates that characterize other scientific fields — they’re engineering problems demanding concrete solutions. CDS Assistant Professor of Psychology and Data Science Grace Lindsay recognized this distinction when she launched “5 Minute Papers on AI for the Planet,” a YouTube channel that distills complex research papers at the intersection of artificial intelligence and environmental science.
The channel emerged from Lindsay’s undergraduate course at CDS, “Machine Learning for Climate Change.” Each week, her students tackle a research paper while Lindsay provides background lectures covering both the machine learning techniques and the domain knowledge — whether that’s energy systems, deforestation patterns, or biodiversity monitoring. “You’re learning things about the world,” Lindsay said. “But then you also can learn about the machine learning side, the technical side.”
Lindsay’s path to climate-focused work wasn’t straightforward. After spending years using artificial neural networks to model the brain’s visual system during her PhD and postdoc, she made a deliberate shift in 2021. The confluence of factors — the pandemic, returning to the US from the UK, becoming a parent — pushed her to reconsider her research priorities. “Can I just be a neuroscientist when there are these big problems in the world?” she asked herself at the time.
Her computational neuroscience background proved surprisingly relevant. Her lab now analyzes satellite imagery for climate change applications, leveraging the same visual processing expertise she developed studying the brain. The methodological overlap between studying vision in neural systems and analyzing satellite images created a natural bridge between her past and present work.
The YouTube format represents a strategic choice based on how people consume information today. Unlike her previous podcast on computational neuroscience, which featured discussion-based episodes, these videos are structured as straightforward five-minute paper walkthroughs. “When you’re talking about something that’s a little bit more engineering, there isn’t that much big discussion to have around it,” Lindsay explained. Each video follows a consistent narrative arc: problem, solution, proof of effectiveness.
The channel’s early videos reflect the range of topics Lindsay plans to cover — from “Using AI to Find Solar Panels from Space” to “How AI Can Help Monitor Bird Species” to “Learning Better Weather Forecasts with AI.” Lindsay deliberately selects papers from major labs tackling significant problems, aiming for broader relevance than the pedagogically constrained selections for her course.
Creating each video requires several hours of work, with the most time-consuming part being digesting the papers themselves — especially when they lack details or cover unfamiliar territory. The editing process adds another layer of complexity, as she incorporates screenshots and diagrams from the papers to illustrate key concepts.
Science communication has always served a dual purpose for Lindsay. Her previous podcast gave her an excuse to read papers outside her direct research area. Her book on computational neuroscience let her explore the field’s history. Now, these videos allow her to stay current with climate-AI research while making that knowledge accessible to others. “I frequently use science communication as an excuse to learn things that I want to learn about,” she said.
The channel’s target audience remains intentionally broad — people with climate expertise seeking to understand AI applications, AI practitioners curious about environmental applications, or anyone interested in how these fields intersect. Lindsay hopes to maintain the interdisciplinary approach that defines her course, covering everything from technical machine learning advances to social science applications.
As video increasingly dominates the online information landscape, Lindsay’s five-minute format offers a middle ground between lengthy academic papers and oversimplified summaries. Each episode delivers technical accuracy without sacrificing accessibility, serving both as standalone educational content and as an entry point into deeper research.
By Stephen Thomas
Have feedback on our content? Help us improve our blog by completing our (super quick) survey.
