The topic of discussion at our November meeting was AI and the Environment. Whilst as a group we clearly love AI and what it can do, we learned through this meeting that AI is contributing to global warming. There were a number of fruitful discussions, and we tried to focus on solutions rather than problems. Mike Newell took on chairing the meeting at short notice. Click here to listen to an audio recording of the meeting.
Penny James shared a video called "Is AI fuelling a New Climate Crisis?" by Stuart Clark, which you can watch on YouTube. YouTube summarises the video as:
"We don’t often think of data as polluting, but what if every scroll, stream, and swipe came with a carbon cost? In this powerful talk, we’re asked to confront the hidden environmental toll of our digital lives - from AI’s massive energy demands to the mountains of e-waste we leave behind. As data centres expand and screen time soars, the speaker challenges us to rethink our relationship with technology. Can we build a future where digital progress doesn’t cost the Earth?"
The meeting focused on the environmental impacts of modern data infrastructure, especially the rapidly growing global demand for data centres. Speakers explored both innovative solutions and the significant risks associated with them. A major part of the conversation centred on proposals to place data centres either underwater (Natick, Highlander, Subsea Cloud) or in space. Underwater facilities could dramatically reduce cooling energy usage and improve reliability, but participants noted serious concerns: warming the surrounding sea, potential ecological harm, and the complexity and cost of maintenance. There was also discussion of the idea of space-based data centres, which might offer constant solar energy and vast capacity but would be limited by extreme launch costs, radiation risks, and the environmental footprint of rocket launches
Participants also highlighted wider environmental challenges linked to conventional land-based data centres, such as their heavy electricity and water consumption, and the fact that many are built in regions already experiencing water scarcity. Examples were given of very large new centres being constructed internationally and locally, raising concerns about competing demands for land and resources.
The group then shifted to consider the positive environmental applications of AI, including wildlife monitoring, smarter farming, improved renewable-energy management, deforestation detection, and the design of more efficient structures. They also reflected on the growing role of AI in education, agriculture and sustainability research. Sustainability includes actions by companies such as Prologis, and Marks and Spencer. However, there was ongoing tension between AI’s potential benefits and its resource-intensive infrastructure, alongside questions about regulation, global responsibility, and whether strategic policy documents from governments e.g. Scotland offer meaningful oversight.
A recurring theme was intergenerational responsibility: younger people may inherit the environmental consequences of today’s technological choices, and participants wondered how younger generations perceive both AI and climate change. The meeting concluded with discussion of travel, pollution, plastic waste in oceans, and the difficulties of achieving effective international regulation.
A. Underwater Data Centres
Potential advantages:
- Dramatically reduced cooling energy requirements
- Increased reliability due to stable temperatures and reduced human interference
- Ability to collocate with offshore renewable energy
- Modular construction enables rapid deployment
Key concerns
- Heat released into the ocean may affect marine ecosystems
- Long-term ecological impact is unknown
- Maintenance is extremely challenging and requires specialist robotics
- Engineering demands for long-term underwater durability
B. Space-Based Data Centres
Possible benefits:
- Continuous access to solar energy
- Vast potential physical capacity
- Protection from Earth-based disasters
Major drawbacks:
- Cooling in the vacuum of space is difficult
- Very high financial and environmental cost of rocket launches
- Radiation and space-debris risks
- Very limited ability to repair or maintain equipment
C. Environmental Impact of Conventional Data Centres
- Currently consume ~1.5% of global electricity; expected to double by 2030
- Heavy use of freshwater for cooling, often in regions already facing water scarcity
- Large land requirements and significant local environmental footprint
D. Positive Environmental Uses of AI
- Wildlife monitoring through automated image and sound recognition
- Precision agriculture: smarter irrigation, fertiliser optimisation, improved yields
- Managing fluctuations in renewable energy production
- Detecting deforestation and environmental damage from satellite imagery
- Generative AI used to improve image quality, create efficient designs, and aid materials discovery
E. Regulatory & Ethical Issues
- Global oversight of AI and data-centre infrastructure is fragmented
- Many guidelines are non-binding and lack enforcement
- Environmental permitting is inconsistent; some operators bypass requirements
F. Broader Environmental Issues Raised
- Plastic pollution harming wildlife
- Pollution from shipping and air travel
- Potential use of data-centre waste heat for desalination
G. Generational Perspectives
- Younger people feel they are inheriting the environmental consequences of older generations
- Schools restrict AI use, but students circumvent controls
- Interest in involving younger speakers in future meetings