SHADE Newsletter 9th January 2025

Welcome to the twenty ninth edition of the SHADE newsletter! 

SHADE is a research hub with a mission to explore issues at the intersection of digital technologies/AI, health and the environment. It is guided by a fundamental question: How should the balance between AI/digital enabled health and planetary health be struck in different areas of the world, and what should be the guiding principles?

The SHADE newsletter comes out every two weeks, bringing you a selection of the latest news, upcoming events, academic publications and podcasts in the SHADE space.

In this newsletter, we highlight the ongoing march of digital technology and AI in healthcare. We take a sweep through the latest on AMR, data centre water footprints, pandemic preparedness and AI’s energy consumption and risk register. We check in on stories covered in 2024 and take a look at hardware turnover, ‘health informed computing’, benchmarks for the effects of climate change and much more.  We hope you enjoy it!

Please tell us what you like, what you don’t like and what you think is missing at [email protected].   

Highlight on

• New year predictions bear out the transformative promise of digital technology and AI in healthcare, each with a different emphasis. Tech EU has a caveat - benefits will mainly accrue to those who can afford it. The BMJ warns against losing the compassion, understanding, and relationships that make medicine uniquely human. TechTarget warns against innovation inequity affecting smaller or rural health organisations and recommends prioritising data quality, governance and high-value use cases and Forbes agrees, highlighting that the future of healthcare isn’t simply about new technologies.

• The evidence base for the benefits of AI in diagnostics grows - most recently in breast and ovarian cancer.

• Meanwhile in the UK the Health Foundation asked the public how they felt about health technologies and data. Positive, on balance, was the response, with concerns echoing the caveats in new year predictions.

News

• As the Guardian reports that the antibiotic emergency ‘could claim 40 million lives in next 25 years’, the importance of achieving Universal Health Coverage targets to combat AMR is demonstrated in this paper from Scientific Reports.

• A round up of recent developments in stories this newsletter has followed over the past year: First up, Microsoft shareholders have rejected a resolution aimed at addressing the company’s links with the fossil fuel industry. Secondly, Jo Lindsay Walton highlights the flawed evidence behind claims that AI will save the planet. Finally, Sasha Luccioni predicts that 2025 will be the year we finally start to grapple with AI’s environmental impacts, helped by tools such as the energy star ratings for AI she is developing.

• Microsoft have released their plan to reduce the water footprint of their data centres with a closed loop, ‘zero water evaporation’ system that results in only a ‘nominal increase in our annual energy usage’.

• The BBC reports on an analysis from the Intergovernmental Platform on Biodiversity and Ecosystem Services (IPBES), warning of the dangers of ignoring the links between biodiversity, water, food, health and climate change.

What we’re listening to

• This December episode of GreenIO examines the rise and fall of Moore’s Law. It looks back at how its rise bought about the rapid turnover in hardware and allowed inefficient software to flourish. Then it considers how the demise of Moore’s Law has led to a drive to make software efficient, and how this could finally mean a slowdown in hardware turnover with all the benefits for the environment that entails. (Just think twice about asking AI to optimise the efficiency of your software).

• In Building Local Power from the Institute for Local Self-Reliance, Dr Shaolei Ren examines the environmental inequity of AI and calls for ‘health-informed computing’.

• As fires rage in Los Angeles, this episode of Bloomberg’s Zero: The Climate Race podcast asks If 1.5C is dead, what happens next?

What we’re reading

• Nature looks at AI’s role in the holy grail of pandemic preparedness - determining what viruses will do next. Still on the pandemics theme, this analysis in Scientific Reports uses COVID-19 as a benchmark for measuring the effects of climate change. Comparing heat related mortality to COVID-19 mortality, it concludes that ‘the consequences of climate change are not as distant as they may seem, and the threat to human lives is comparable to the profound global impact of the COVID-19 pandemic’.

• The Unpaid Toll: Quantifying the Public Health Impact of AI in Computers and Society. Hear more from one of the paper’s authors, Dr Shaolei Ren, in ‘what we’re listening to’ above.

• Following on from this, a flurry of articles on AI and its environmental impact give insights into a complex and evolving debate: First up the MIT Technology Review says AI’s emissions are about to skyrocket even further as multimodal models comes out of the research phase. In a similar vein, Professor Rabih Bashroush uses OpenAI’s latest O3 AI model to calculate how AI’s energy consumption could ramp up to consume nearly the entire current global power generation capacity. Various technical solutions are proposed to address this energy consumption. Firstly, Nature reports on a breakthrough in light management technology that could allow future mass manufacturing of photonic chips to address soaring energy consumption in data centres. Secondly, this paper has made waves at a recent AI conference with its suggestion that the next generation of AI could live in hardware rather than software and in doing so address environmental concerns. Thirdly, an old idea based on the laws of thermodynamics has been resuscitated and reversible computing escapes the lab in 2025. Finally a possible non technical solution, as Nature reports that the AI revolution is running out of data and consequently AI in the form of big all-purpose large language models (LLMs) might make way for smaller, more specialised models.

• This paper in Science analyses the risks posed by mirror organisms, which have a different ‘handedness’ to those occurring naturally. Technological advances - including AI - mean the synthesis of mirror bacteria could be possible with the next decade or so. The authors conclude that the risks are such that the creation of the mirror organisms should be prevented. They call for critical engagement with their analysis and propose a path forward.

• Sticking with AI risks, the MIT AI Risk Repository added 13 new risk frameworks in December 2024. In 2025, to improve coordination on AI risks, it is launching a new initiative to document and evaluate organizational and institutional responses to high-priority AI risks.

Events

• Our Planet, Our Health: 2025 Climate Action Convention is happening in Washington DC and online, March 1st to 4th.

Opportunities

• Three postdoctoral positions at the University of Oslo for research on topics related to climate and health modeling have been announced by the HISP Centre: First, A sociotechnical approach to predictive modeling of climate health. Second, Machine learning on climate, health and biological data and third, Quality and sustainability of machine learning software for climate change. All are available through the DSTrain postdoctoral program.

We hope you have enjoyed this newsletter. If it has been forwarded to you, and you would like to receive future editions, you can subscribe here