Balancing the Potential of Healthcare AI with Environmental Impact

The healthcare industry is increasingly adopting artificial intelligence (AI) for various tasks, such as responding to patient inquiries and streamlining operations. A recent study conducted by Cornell University highlights the importance of integrating sustainability considerations into the development and deployment of AI systems in healthcare.

The research introduces the framework called Sustainably Advancing Health AI (SAHAI), which aims to optimize the energy consumption and reduce emissions associated with AI applications in healthcare settings. This approach takes into account greenhouse gas emissions from AI-driven patient communication, water usage for cooling hardware in data centers, and potential scenarios that could affect a health system's environmental footprint.

Dr. Chethan Sarabu, director of clinical innovation at Cornell Tech, emphasizes the need for healthcare organizations and technology developers to evaluate various factors—such as energy use and water cooling requirements—when deploying AI tools. He points out that small improvements in model accuracy could result in disproportionate increases in emissions, underscoring the importance of making environmentally conscious decisions early in the system design phase.

An illustrative example from the study estimates that operating an AI-powered messaging system for a year, handling approximately 50 messages per day for 3,000 physicians, could produce around 48,000 kilograms of CO2—equivalent to about 2,300 tree-years of carbon absorption. This estimate was based on using a lightweight generative transformer model, which consumes less computing power than larger models but may be less capable for complex tasks.

Udit Gupta, an expert in computer architecture at Cornell, indicates that sustainability efforts need not compromise performance. For example, choosing data centers powered by renewable energy sources can significantly decrease operational emissions. The researchers argue that proactively considering environmental impacts during the planning stages of AI implementation is more effective than retrofitting systems later.

Despite the benefits of AI in easing healthcare provider workloads—projected to grow into a $187 billion industry—its energy demands raise concerns about sustainability. The study stresses that high-resource, high-income healthcare systems in developed countries are major contributors to AI-related emissions. Therefore, addressing these environmental costs is crucial, particularly because climate change disproportionately affects vulnerable populations.

The authors conclude that there is a crucial window of opportunity to guide AI integration in healthcare with environmental responsibility in mind. They advocate for strategic investments in renewable energy and smarter data center placements, emphasizing that early action can lead to more sustainable health technologies in the future.

Source: https://medicalxpress.com/news/2025-09-health-ai-carbon.html