AI x Climate x Health | 1/21/24

Climate Health Innovators! Embark on a bi-weekly journey where we explore groundbreaking ideas, pivotal news, and exciting opportunities designed for anyone interested in championing climate action in healthcare. If you know someone who would benefit from CHILLing with us — clinicians, builders, or funders — tell them to subscribe here. 📬

Deep dive: AI x Climate x Health

Artificial intelligence (AI) promises to transform every aspect of our society from how we work, communicate, all the way to tackle some of our biggest challenges across healthcare to climate change. Simultaneously, there are great challenges from algorithmic bias to the rapidly growing environmental footprint of training and using AI models with growing scale and alacrity. 


AI encompasses a melange of technologies from those that make robust predictions to generating novel text based on contextual understanding. This report, ‘Accelerating Climate Action with AI,’ by Google and BCG provides a thoughtful breakdown of the different flavors of AI and how they might apply to climate change.

Explanations of different types of AI and examples of how they may help with climate related issues. (Source: Google BCG report ‘Accelerating Climate Action with AI’)

AI’s ability to simultaneously comb across complex health and environmental data streams, to then forecast outcomes in a way that bolsters our preventive and adaptive health strategies holds massive promise. Across climate prediction, disaster response, pollution monitoring, and disease outbreak surveillance, AI models provide a powerful tactical capacity that can help healthcare organizations and public agencies better optimize increasingly strained resources. 


As AI models are dependent on vast swaths of data for learning patterns and rules, they are only as rich as their underlying training data, and can often amplify encoded biases. When using tools like GPT-4 in healthcare, researchers have found gender or race-skewed diagnostic accuracy based on biases preexisting in the model’s training data. Independent groups such as the Algorithmic Justice League and the DAIR-institute have been sounding this alarm and offering strategies on how to ensure that AI is equitable and accountable. Healthcare coalitions such as CHAI, VALID-AI, and the US Government have been calling for more guidance and standards that can help safely deploy AI in healthcare.

The Climate Footprint of AI

According to a recent analysis highlighted in the New York Times, within the next 3 years, AI servers are projected to use the same amount of energy as a country such as Sweden or Argentina, or about 0.5 percent of the world’s total electricity use. Most of our understanding of the environmental footprint of AI comes from the costs around training new models, when a significant amount of electricity and water are needed to power and cool the graphical processing units (GPUs) during the development process. We are also learning more about the cost of using models, and one recent study by AI startup Hugging Face found that creating images vs text was responsible for significantly more carbon emissions, generating 1,000 images is similar to driving 4.1 miles in a gasoline-powered car. This carbon consideration is prompting AI researchers and implementers to consider methods to reduce the footprint of AI with many great suggestions included in this Harvard Business review article from choosing energy efficient models to optimizing when they are used. 


Healthcare organizations are simultaneously signing climate pledges to reduce their energy footprint while also racing to deploy energy-intensive AI tools across the enterprise. At the recent AMIA Annual Symposium during a workshop on Climate Health Informatics led by CHILL co-founders Chethan Sarabu and Stefano Leitner, participants used publicly available data to estimate the energy footprint of using an LLM to assist clinicians in sending messages to patients. Based on best estimates using publicly available data from LLM companies, an AI-assisted message system may add about 2% additional energy use to a medium-sized hospital. Healthcare leadership needs to consider how to best align climate pledges with the increasing energy requirements of AI and develop joint strategies with ideas such as ensuring that new energy expenditures are powered by renewable sources. Ultimately, with any new technology, there are significant benefits but also risks and externalities to consider, particularly so with the promise and footprint that AI offers at the climate and health nexus.

Other news

  • A large-scale, first-of-its-kind study from Curtin University in Australia revealed a possible link between extreme bioclimatic exposure during pregnancy and babies’ birth weights for gestational age. (Technology Networks Applied Sciences)

  • Harvard’s Kari Nadeau gave a TEDxBoston talk where she quantified the value of climate investments to health. “If you were to invest $1 in climate change resilience solutions you could get at least $6 out in health benefits.” (Harvard T.H. Chan School of Public Health)

  • Researchers are studying how extreme weather and rising temperatures can encourage the spread of drug-resistant infections. (Nature)

  • Dr. Vanessa Kerry argues that even as philanthropy pledges billions toward climate adaptation and the climate-health nexus, what really matters is how and to whom those dollars are being deployed — and it needs to be "catalytic, flexible, and fast.” (Inside Philanthropy)

Events & opportunities

  • The Medical Society Consortium on Climate and Health hosts their meeting in Arlington, Virginia, on February 11-12, with a visit to Congress on February 13. Register.

  • The 2024 Planetary Health Summit and 6th Annual Meeting will be on April 16-19 uniting scientists, educators, policymakers, and industry leaders to shape a sustainable and resilient future. Register.

  • Navigating NetZero Bootcamp by the American College of Healthcare Executives (ACHE) March 25 - 28 in Chicago. Register

  • Join the Climate Advocacy Lab on January 30 starting at 2PM ET / 11AM PT in an interactive webinar on “How to lead as an evidence-based advocate.” Register.

  • Join an interactive event on January 30 to learn more about parenting in the climate crisis. Register.

  • Join Climate Health Now’s Inaugural Discussion Club on January 31 starting at 3:30 PM ET / 12:30PM PT! They will be discussing the documentary "Kiss the Ground," which focuses on regenerative agriculture as a climate solution. Register.

B