Page 288 - GHES-3-3

Global Health Economics and
Sustainability
Carbon footprint of smartphones in healthcare

Another contributing factor could be the complexity of the CO emissions of their smartphones. Our research
2
interpreting carbon footprint data. Environmental reports demonstrates a critical need to transform how healthcare
often present emissions data in varying formats or as part professionals use smartphones in their daily practice.
of broader sustainability metrics, which may confuse the The smartphones represent the central component of the
LLMs or lead to misinterpretation. clinical ICT ecosystem.
Prompt specificity also plays a critical role in the Present research suggests that the healthcare
accuracy of AI responses. While the question posed to the information system will become increasingly resource-
LLMs: “What is the average carbon footprint of the entire intensive as healthcare facilities integrate internet-of-things
lifecycle of (smartphone model)?” was straightforward, technologies into smart hospital systems, for example,
additional contextual prompts or follow-up queries might medical monitoring systems and wearable health devices
have improved the reliability of the responses. For example, (Kögler et al., 2024), connected diagnostic equipment
asking the models to clarify their sources or specify (Marosi et al., 2018), and other emerging digital healthcare
whether the information is directly cited from corporate solutions (Paulick et al., 2022). Table 5 showcases the power
reports could have reduced discrepancies. consumption and energy usage through the smartphone’s
display. It shows the average power consumption and
These findings suggest that while LLMs have the power budget through the respective smartphone setting
potential to broaden access to environmental data, their (Kögler et al., 2024).
accuracy is constrained by the availability and clarity
of the training data sets. Moving forward, improving We conducted an anonymous internet survey over
the integration between AI tools and standardized, 1 week by creating a post on Reddit. We asked three
verified datasets is essential for enhancing the reliability questions about smartphone purchasing decisions. The 78
or veracity of data (Schneider et al., 2024). In addition, anonymized responses showed that the majority (67/78,
developing mechanisms for LLMs to communicate 86%) did not consider carbon footprint when purchasing
uncertainties or data gaps transparently could help a smartphone. While this represents a small study, the
mitigate the risks associated with overestimation or findings align with anecdotal evidence gathered by one
misinterpretation. of our authors (Gloria Wu), who informally surveyed
over 50 physicians. These healthcare professionals also
At present, the field of AI-mediated LLMs uses overwhelmingly reported not considering the carbon
foundational data sets as a form of “grounding” to prevent footprint in their smartphone purchasing decisions.
hallucinations or “guessing” by the models. The field of Notably, 55% of the physicians in the Internet survey
AI-mediated LLMs and their capabilities are expanding as were unaware that smartphones have a carbon footprint
of this writing (Schneider et al., 2024). (personal communication) (Wu & Paliath-Pathiyal, 2024).

The lack of globally accepted standards for smartphone The public is not aware of the carbon footprint that is
CO₂ emissions hampers the future manufacturing being produced through smartphones. Based on the
specification of the LLMs. The accepted Paris Accord analyses of current usage patterns and projected growth,
does not mention smartphones (Wu, 2024). There is an healthcare institutions should consider sustainable
accepted international GHG Protocol. The United States device management protocols and energy-efficient usage
formally withdrew from the Paris Accord in November techniques that will enable clinical professionals to reduce
2020, rejoined under President Biden in February 2021, the carbon footprint of their digital tools while maintaining
and withdrew again under President Trump on January 20, high-quality patient care.
2025. Furthermore, the GHG has no policing mechanism, Our findings underscore the need for a standardized and
and the participating nations are following these guidelines authoritative source of smartphone carbon emissions data.
in a voluntary manner (World Resources Institute, 2004). The variability in reported emissions across LLMs obscures
The lack of interest in the general public regarding CO public understanding of the true environmental impact
2
emissions of smartphones means that fewer queries are of smartphone production and usage. Policymakers and
directed to AI-LLMs, which is another reason that they environmental advocates lack a comprehensive, accessible,
are not “trained” to respond to occasional queries about and verifiable dataset to assess the true ecological footprint
smartphone CO emissions and the related standards. In an of smartphones. Sustainable management strategies can
2
unpublished survey of 100 respondents, we found that 98% only occur with data on all medical devices and their
of the respondents had a smartphone, but none of them carbon footprint in the healthcare setting.
knew about their CO emissions (Wu, 2024). Thus, there AI chatbot queries on smartphones are increasing from
2
is a lack of knowledge among the general public about 100 million users of ChatGPT in Jan 2023 to 400 million

Volume 3 Issue 3 (2025) 280 https://doi.org/10.36922/ghes.8359