This series is a brief look at the content within Harnessing AI for Environmental Justice, a report WAO developed for Friends of the Earth, funded by Mozilla. Read or download the full report here.
Check out Part 1, Understanding Predominant Narratives in AI. In this second part, we look at the systemic complexities around environmentalism and technology.
- Understanding Predominant Narratives in AI
- Systemic complexities with AI
- Starting principles for the ethical use of AI (coming soon!)
- How to be an Activist in a World of AI (coming soon!)
When viewed with all of the facts, nothing is simple. The same is true with AI, and everything else related to environmental justice and digital rights. As a result, we need to be aware of complexity.
The climate crisis disproportionately affects low-income communities, women and marginalised communities that are already affected by systemic inequalities. (Kazansky, 2022). We know that ‘black-box’ algorithms, a lack of regulation of corporate conglomerates, and obscured climate impact reporting is common within the tech industry. (Kazansky, 2022). Companies are not obligated to disclose essential information in the proprietary nature of their AI development, including labour practices in the AI supply chain and procurement of materials or rare minerals. (Kazansky, 2022). Digital rights activists tell us how data privacy and surveillance is affecting our most vulnerable communities as well as wider society.
These issues are systemic, touch many different policy points, and are best addressed through thoughtful coalitions and the important, painstaking work of activism.
Climate and Energy
AI is being used to address the climate emergency and reduce energy consumption in various ways. For example, the UN talks about improving prediction modelling for how climate change is impacting our societies using machine learning and predictive AI. (United Nations, 2023). Other companies are using similar technologies for helping prevent wildfires, promoting reforestation or improving early warning systems. (Climate One, 2024).
However, none of these are using generative AI for positive climate action and to tackle the rising energy consumption required by new data centres.
Journalists from across the environmental and tech sectors repeatedly emphasise the significant energy consumption associated with generative AI models like ChatGPT. They point out that the energy and water usage to run and cool data centres is growing exponentially because of AI, and that by 2027 the sector will have the same annual energy demand as the Netherlands. (Vincent, 2023). However, it’s important to understand that, just as when we compare one country’s emissions with the rest of the world, data centres are responsible for a small percentage of total global energy-related emissions.
This does not mean that there is nothing to concern us here. As we have not yet transitioned to global renewable energy, increased energy usage incentivises the use of fossil fuels. (Kazansky, 2022). Indeed fossil fuels are still responsible for over 80% of our global energy needs. (Ritchie, 2024). The energy mix in some places is increasingly renewable, whereas other places are almost entirely dependent on approaches emitting high levels of CO2.
Nature and Environment
AI can be used to help in everything from conservation efforts with endangered species (Wildlife.AI, 2024) through to dealing with controlling pests such as the Desert Locust. (PlantVillage, 2024). In these cases, AI usually takes the form of object detection and tracking, computer vision, and machine classification. This is often combined with cloud computing, analytics and satellite intelligence to provide insights for farmers, activists, and conservations to take action.
Generative AI is not central to these efforts, but the energy being used to generate synthetic text, images, and video is a mainstream issue. Related is the issue of resource consumption, which tends to receive less attention. This is particularly true of freshwater usage, as data centres use water for cooling, and the explosion of generative AI technologies has exacerbated freshwater water scarcity. The trend shows no signs of slowing. (Li, Pengfei, et al., 2023).
Local communities in areas where data centres are built deal with a variety of ecological and social issues due to this resource consumption. Human communities, farm lands, natural areas and biotopes suffer as Big Tech works to hide its water usage. (Smith & Adams, 2024). Data centres are competing with these local communities and ecosystems for scarce water resources, especially in regions already experiencing drought. (HuggingFace, 2024). This competition puts a further strain on water availability and poses a threat to biodiversity in these vulnerable areas.
Clean technology and innovative, cyclical solutions in resourcing is critical in the protection of biodiversity.
Rights and Justice
Machine learning, image recognition systems and predictive analytics can be used to work on a number of issues related to the UN’s Sustainable Development Goals (SDGs). For example, issues relating to hunger, education, health and well-being can be addressed in part by AI technologies. We are beginning to see innovative uses of AI such as helping to identify or predict illnesses as well as showing us ways to reduce food waste.
However, good examples of how generative AI are supporting such progressive initiatives are hard to come by. When it comes to rights and justice, it is much easier to unpack how generative AI is causing harm. Alongside the more visible issue of resource use, a slew of discriminatory and exploitative practices are made worse by the AI boom.
The waste and pollution, ecosystem collapse and worker exploitation associated with AI advancements disproportionately impact the communities who are already most vulnerable to the climate crisis.
Civil society organisations and communities play an important role in ensuring that extractive practices, in the context of resource extraction, labour and further colonialism, are pivotal in shaping AI development.
There’s more to read in the full report. Read or download it here.
Technology has changed so much about our world. Now, AI is changing things again and at hyperspeed. We work to understand its potential implications for our communities, learners and programmes. Do you need help with AI Literacies, strategy, or storytelling? Get in touch!
References
Kazansky, B., et al. (2022). ‘At the Confluence of Digital Rights and Climate & Environmental justice: A landscape review’. Engine Room. Available at: https://engn.it/climatejusticedigitalrights (Accessed: 24 October 2024)
United Nations (2023). ‘Explainer: How AI helps combat climate change’. UN News, 3rd November. Available at: https://news.un.org/en/story/2023/11/1143187 (Accessed: 24 October 2024).
Climate One (2024) ‘Artificial Intelligence, Real Climate Impacts’ [podcast], 19th April. Available at: https://www.climateone.org/audio/artificial-intelligence-real-climate-impacts (Accessed: 24 October 2024).
Ritchie, H., Rosado, P., & Roser, M. (2024). Renewable Energy. Our World in Data. Available at: https://ourworldindata.org/renewable-energy (Accessed: 28 November 2024)
Wildlife.AI — Using artificial intelligence to accelerate conservation. (n.d.). Wildlife.AI. Available at: https://wildlife.ai/ (Accessed: 11 December 2024).
Plantvillage. Penn State University (n.d.). Available at: https://plantvillage.psu.edu (Accessed: 11 December 2024).
Li, Pengfei, et al. (2023) Making AI Less “Thirsty”: Uncovering and Addressing the Secret Water Footprint of AI Models. Available at: https://arxiv.org/pdf/2304.03271 (Accessed: 20 November 2024)
Smith, H. & Adams, C. (2024) ‘Report: Thinking about using AI?’. The Green Web Foundation. Available at: https://www.thegreenwebfoundation.org/publications/report-ai-environmental-impact/ (Accessed: 24 October 2024).
Hugging Face (2024) The Environmental Impacts of AI. Available at: https://huggingface.co/blog/sasha/ai-environment-primer (Accessed: 24 October 2024).