.png)
Introduction
Maya sits at a small white table in her kitchen sipping her coffee before she logs on for the day. Usually, her mind drifts to the model she’s working on to predict chronic kidney disease, but this morning her attention is captured by headlines of how wildfires have been linked with tens of thousands of premature deaths in California, USA within a decade and they’re becoming more intense due to climate change. She was familiar with the heartbreak that can accompany wildfires after one nearly burned down her hometown, but she hadn’t realized the severity of the health effects. She started digging for more links between climate and health and was astonished to find a mass of headlines and articles – extreme heat affecting chronic conditions, pollution worsening asthma, droughts contributing to water insecurity – the list goes on. She had always been interested in health and felt her work was meaningful, but now she wonders about the role AI has to play in the intersection of climate and health.
While AI has the potential to enable healthcare systems to be more sustainable and adapt to climate change, AI solutions can also negatively impact the environment if their footprint is not thoughtfully addressed. While we’d do well to incorporate environmental considerations into responsible AI across industries, it’s particularly pertinent in healthcare where health is paramount. With this in mind, the below offers a starting point to understand how you can reduce the environmental cost of your technology, support healthcare systems’ sustainability, and enable health sector climate adaption.
Mitigating AI’s environmental impact
Despite the potential benefits AI solutions can offer to patients, providers, and even reducing the environmental toll of healthcare systems, they can have their own negative influence on the environment. For example, energy-intensive training and data centers can contribute to greenhouse gas emissions, the hardware to power AI requires the extraction of rare earth metals, and resultant electronic waste is not always safely recycled (Katirai). As a result, we must ensure we are considering the toll our technology takes on the earth and how to mitigate it. While there are multiple stages of the technology lifecycle to consider, Hugging Face provides an example of a company researching model efficiency via their Energy Star AI Project to enable reducing energy used. We can each begin to explore the footprint of our technology by asking:
How might we begin to understand and measure the environmental impact of our technology across its lifecycle and conduct a lifecycle assessment? Can we perform ongoing monitoring to enable continually identifying opportunities for improvement?
How might we reduce the footprint of our technology across its lifecycle and design for this from the outset (e.g., use renewable energy, practice responsible data management, enable reducing or safely recycling e-waste)?
How might we share our measurements and impact to promote a culture of sustainability in healthcare and normalize environmental considerations as a tenet of responsible AI?
Enabling sustainable healthcare systems
There are multiple opportunities for AI to help reduce the toll healthcare takes on the earth, such as optimizing workflows, equipment usage, and resource allocation (Ueda et al.). Enhancing telemedicine and automating diagnostics are two of the most readily available examples – these can help prevent unnecessary office visits, which can reduce greenhouse gas emissions from commuting, and enable more efficient interventions. When AI solutions support mitigating healthcare’s environmental impact, this can in turn help prevent worse health outcomes due to environmental degradation and climate change. To understand your technology’s potential to enable mitigation, you can begin by asking:
How might my technology enable healthcare systems to become more sustainable (e.g., reduce greenhouse gas emissions, minimize waste)? How might I adapt my technology or design a new tool to do so?
How might I measure the impact of my technology on healthcare systems reducing their footprint? Can we continuously monitor and elevate the mitigation it enables?
How might we strive to ensure the mitigation enabled by our technology outweighs its environmental cost?
Supporting healthcare’s climate adaptation
Additionally, AI has the potential to support healthcare systems’ climate adaptation. This refers to reducing the sector’s vulnerability to and protecting community health in anticipation of the impacts of climate change (e.g., increased natural disasters, additional extreme weather, worsened water or food insecurity). Examples of how AI tools can enable resilience include prediction of and preparation for climate shocks, extreme weather, and disease outbreaks of climate-sensitive conditions. These early warnings can then inform resource planning and proactively supporting patients. For example, Fraym mapped risks of Ebola outbreaks informed by ecology and human behaviors to enable allocating resources (e.g., vaccines, personal protective equipment) to the communities with the highest level of risk. To explore how your technology can enable healthcare systems to be climate resilient, you can begin by inquiring about the below:
How might this technology enable healthcare systems’ climate adaptation? Can I design for this from the outset or adjust my AI solution to enable resilience?
Which stakeholders and perspectives do I need to consider to ensure my technology effectively enables climate adaptation? How might I collaborate with them or invite their input?
How will I measure the impact of this technology on climate resilience, ensure it has the intended influence, and continually make it more effective?
Conclusion
Not only can we mitigate environmental harm from AI tools and healthcare to reduce risks to both environmental and human health, we can even leverage AI to enable healthcare systems’ climate adaptation. While these environmental considerations are particularly pertinent to healthcare given the link between climate and health, anyone in any industry can engage with the questions above, promote environmental considerations as a key element of responsible AI, and ensure your technology is helping more than harming the environment.
Alexandra Crew is a health and AI ethics expert with publications in peer-reviewed journals and industry publications. She is an AI Ethics Expert at Ethical Intelligence. She works in healthcare full-time and is part of the AI Strategy Working Group informing the path forward for emerging technologies at her employer. She is also part of the Planetary Health Alliance’s Balancing Business Working Group focused on mainstreaming planetary health and sustainability into business and industry.