We Added New Cases to the Observatory of Planetary Justice Impacts of AI!
A new batch of case studies has been added to our Observatory of Planetary Justice Impacts of AI! This is one of our core initiatives, where we map real-world examples of the planetary justice impacts of the AI supply chain. It is a continuous, adaptive effort to collect data, to track cases and investigations, and see what patterns emerge.
The effects that AI is having on our planet seem to be more stark in the Global Majority - part of our mission is to map how the AI supply chain follows colonialist patterns in extracting resources from the Majority to benefit the Global Minority. However, naturally, the Global Majority and Minority are not homogenous entities, and extractivist patterns exist within and across them in a complex web, affecting marginalized people at the intersections of multiple geographies and power dynamics. While so far we have included a few studies of Disposal/End-of-life in Western and Central Africa and in Southeast Asia, we have now seen how toxic waste from the AI supply chain is present in Europe as well, affecting those who live around it and draw from nearby water resources. Phoebe Weston reported for the Guardian on a disused mine in Alsace containing “mercury, arsenic and other heavy metals as well as cyanide and residues from household waste incinerators.
The City of Memphis. Source: Unsplashed.
These power dynamics are starkly evident elsewhere in the Global Minority as well. Data centres are increasingly coming under scrutiny in the US, where a great majority of them is located. Reports from the nonprofit More Perfect Union investigate how life is impacted for people who live very close to data centres, in Georgia and Tennessee. Residents report higher electricity and water bills, effectively subsidising profits for the tech companies that have built these data centres.
Reporting is increasing on xAI’s presence in Memphis, Tennessee, where Elon Musk is accused of building “as many as 35 methane-burning turbines clustered around the factory buildings that are entirely unpermitted,” reports Willy Blackmore. Emissions from the Colossus facility include ground-level ozone, which is an asthma-causing pollutant, as well as formaldehyde, a known carcinogen. These are concentrated in the predominantly Black neighbourhood of Boxtown, which seems to exemplify how the costs of the AI industry are often paid by marginalized communities, even within the Global Minority. Shakil Khan also reports that the illegal turbines “which burn fossil fuels 24/7, emit nitrogen oxides (NOx), carbon monoxide, particulate matter, and other harmful pollutants.” These marginalised communities are organising and resisting their exploitation, calling on their local representatives to deny the permits xAI is belatedly applying for.
We have seen increased global reporting on data centres, which of course play a key role both in the Model Training and Model Deployment stages of the AI Supply Chain. Where Cloud Meets Cement, a new report from The Maybe, follows 5 cases across the world in which data centre construction has met resistance from the community, highlighting successful strategies in Chile, the USA, the Netherlands, South Africa, and Mexico. The study highlights the geopolitical nature of companies’ choices of location to build data centres, as well as its consequences for labour, energy and water consumption in a resource-strained landscape. It also highlights the lack of transparency on companies’ and governments’ part, so that advocacy groups are left to collect the data and lead the investigations on how they are impacted by these hyperscale data centres. The new research on Querétaro, Mexico, adds to the body of work already included in the Observatory, mapping how water consumption is increasing difficulties for indigenous communities to access water supply.
Ekurhuleni, South Africa. Planet Labs. Licensed under the Creative Commons Attribution-Share Alike 4.0 International license.
A Microsoft data centre is also being investigated in Mekaguda - a village in Telangana, India - where the local community claims that “Microsoft has illegally occupied land beyond the boundaries of its property and dumped industrial waste into a nearby lake,” as reported by Rest of the World.
From these cases, we continue to see how the stress on the water and electricity grids is causing dire consequences for communities around data centres, and how the lack of transparency from global corporations prevents such communities as well as the wider public from accessing the real figures of the impact of the AI supply chain. Governments are jumping into building more and more infrastructure to support the growth of the AI industry; however, without accurate data (both qualitative and quantitative), we are unable to see exactly how it will be damaging planetary human and more-than-human ecosystems as the industry expands. Still, it is clear that the damage is happening. As we continue to gather data and evidence, we hope to contribute to building a bigger picture of the impacts of AI on the communities it is affecting.
We want to emphasise that the Observatory is a collective repository, and that we are always looking for submissions of case studies. This is a community-led, adaptive data collection effort. If you come across case studies similar to these, please submit them here!