You might have noticed a new collection of images from Lone Thomasky and Bits & Bäume in the library. These images expose the realities behind the often perceived “clean”, “slick”, and “efficient” AI supply chains. Their collection shows the breakdown of environments and depletion of natural resources caused by the rapid expansion of data centres and mining of critical resources (like lithium, cobalt, and rare earth elements) which underpin (generative) AI.
One of their images, ‘Distorted Lake Trees’ has been selected as the iconic cover image of the Montreal AI Ethics Institute 2025 State of AI Report. Below, we unpack some of the images in the Bits & Bäume collection and why they make a fitting partner organisation to Better Images of AI, particularly since our visuals aim to better communicate about the implications of AI on society, the environment forms a huge part of this.
DOWNLOAD THE IMAGE COLLECTION HERE
The hardware that we manufacture to underpin AI technologies and the data centres that power AI are responsible for an ever-increasing share of global electricity and water consumption, as well as being both habitat and land-intensive. Lone Thomasky and Bits & Bäume’s image collection features a series of distorted photographic images of landscapes to visualise the ecologies behind AI’s supply chain.
AI technologies do not merely rely on “digital clouds”, “they are also physical territories anchored in mines, exploited workers, and degraded environmental ecosystems”. Frequently, visual representations of AI mask the underlying realities of how these technologies are built, overwhelmingly dependent on infrastructure, communities, and resources of the Global South.
These images are a valuable contribution to the Better Images of AI library to counter the sleek aesthetics, magical metaphors, and luminous hardware which are often used to represent AI. Instead, these visuals help us move towards more grounded understandings that connect AI technologies with their material and environmental foundations. As such, using these images, we can improve public understanding of what is at stake when AI technologies are used, invested in, and promoted as a source for eternal good in society.
Lone Thomasky & Bits&Bäume / Better Images of AI / CC BY 4.0
How (and why) were the images created?
Bits & Bäume was founded in 2017 by a coalition of critical tech folks, hackers, eco activists, researchers and climate justice advocates out of the conviction that digitalisation and ecological change must be reconciled. Bits & Bäume is formed and supported by NGOs and organisations across the environmental, ethics, and internet policy landscape. Some of these include the TU Berlin Ethics Lab, Chaos Computer Club, the Young Friends of the Earth Germany (BUND), German Trade Union Confederation, Institute for Ecological Economic Research, and Wikimedia Germany. Their collective activities include hosting conferences to advance academic and civil society discussions, policy work to advocate for government changes, as well as research on digital transformation and sustainability to inform law makers.
Bits & Bäume’s first conference took place in 2018 and brought together 2,000 people from diverse backgrounds and communities. After some smaller ones, the next large conference followed in 2022 –you can find their published conference proceedings here. The images in Bits & Baume’s proceedings may look familiar, given that this was the original purpose that the images were created before being donated to the library. Having already known Lone Thomasky, Bits & Bäume worked with her to create images for the proceedings report that reflected the interdisciplinary contributions to the conference. Visualising their research was important to reach new audiences; a lot of the issues that Bits & Bäume work on are very complex, so choosing to communicate via narratives, images, and aesthetics make their topics more accessible.
Bits & Bäume’s conference proceedings on ‘Digitalisation and Sustainability’
The importance of visualising the material realities of AI development
AI relies on physical infrastructures like data centres to house the specific IT architecture, like servers and microchips, that are needed to train, deploy, and sustain AI applications. The number of data centres has been growing for a number of years, but the AI hype and increased investment into the technology are leading to the mass expansion of data centres, particularly concentrated in the Global South.
Just recently, Google announced that it will invest £11.29 billion to build a data centre in southern India’s Andhra Pradesh state which will contain servers, storage systems, and network equipment along with the necessary power and cooling systems to operate them. Adopting a decolonial lens, Khan argues that these takeovers of land, raw materials and exploitation of labour mimic historical European colonialism. However, instead of just blatant brutal force, the power of persuasion in the need for technological innovation and masking of exploitation create the same imbalances of power that have long exploited Global South communities.
The proliferation of data centres are huge consumers of water which are needed to cool hot servers but also needed indirectly for producing electricity to power these facilities. Since electricity is more costly than water for data centres, companies choose to build data centres where there is cheap power, but this is often in areas where water is already a scarce resource. Bloomberg News found that around ⅔ of new data centres built or in development since 2022 in the US are in places already facing high levels of water stress. For instance, the Large Language Model (LLM) LLaMA-3, developed by Meta, consumed 22 million litres of water over 97 days. This is equivalent to the amount of water an average person in England and Wales would use over the course of 424 years.
Lone Thomasky & Bits&Bäume / Better Images of AI / CC BY 4.0
Data centres also produce electronic waste, which often contains hazardous toxic substances, like mercury and lead, which pose serious consequences to humans and non-humans. ‘Distorted fish school’ in Lone Thomasky and Bits&Bäume’s collection points to one of the lesser considered impacts of AI on our aquatic ecosystems from contamination and pollution.
Beyond this, AI development relies on critical minerals and rare elements with unique magnetic and luminescent properties to create the microchips that are required for the processing activities required for faster, more efficient, and higher performing processing. One example is the Kolwezi copper and cobalt mine in the Democratic Republic of Congo, where Amnesty International documented forced evictions of entire communities as companies expand unsustainable mining operations for technological advancement. The mining of rare earth minerals generates large volumes of toxic and radioactive material which leave land uninhabitable. ‘Distorted Sand Mine’ in Lone Thomasky and Bits&Bäume’s collection illustrates the destructive mining and extraction practices that lurk behind the “stylish, clean, and lightweight” appearance of modern AI tools.
Lone Thomasky & Bits&Bäume / Better Images of AI / CC BY 4.0
Distortion as a Visual Metaphor for AI’s Environmental Destruction
Lone Thomasky & Bits&Bäume / Better Images of AI / CC BY 4.0
Thomasky’s images all feature visual distortion, like bands, pixel shifts and fragmented overlays that disrupt the otherwise natural landscapes of trees, sand mines, and dandelions. The distortions visualise how digital infrastructures, driven by the ideology of progress and innovation, fragment both ecosystems and our perception of them. The juxtaposition between the forest greens and aquatic fish which are fractured by geometric interference draw attention to the mining, energy use, deforestation, and waste, all in the name of technological “innovation”.
A call for greater public understanding and policy shifts about the environmental impact of AI
“The ease-of-use of generative AI interfaces and the lack of information about the environmental impacts of my actions means that, as a user, I don’t have much incentive to cut back on my use of generative AI.” – Bashir (2025)
The visuals in the Better Images of AI library serve many purposes, but for this striking collection, we hope that they’ll improve public understanding of the realities of AI’s supply chain. By revealing the environmental and material realities that underpin AI, we hope to encourage more informed decision-making about whether embedding AI within infrastructures or promoting its use for ‘social good’ is truly desirable.
By countering the clean aesthetics of typical AI imagery with distortion, these visuals invite us to think more critically about the costs of AI, particularly generative AI, which consumes substantially more natural resources than many alternative human methods or technologies. This is not only in their design and development, but even when models are deployed and used, the computing hardware that performs these operations consumes energy. For example, researchers have estimated that a ChatGPT query consumes about 5x more energy than when a simple web search is used.
Beyond improving public understanding, the use of these images in the library can also support wider advocacy efforts that Bits & Bäume have already been spearheading. The community’s 2022 conference included a set of demands aimed at changing government policy on responsible digitalisation. One of their main demands focuses on “digitalisation within planetary limits”. This concept calls for technological progress to align with the standards of climate protection, resource conservation, and biodiversity preservation. In another piece of work, Steig et al counter reductionist understandings of sustainability which narrow the policy space to optimisation and incremental solutionism – running in contradiction to sustainable futures.
“Hackers, researchers, eco and climate justice activists alike are fighting for a livable planet for everyone and for collective freedom, because without clean water you cannot write free software and without democratic digital infrastructures you cannot enjoy healthy forests. And with people globally suffering for our way of life, no one can be happy. Therefore we have to join forces to reflect and act together. Those images contain the joint critical spirit of this endeavour.”
– Rainer Rehak, co-chair of the NGO Forum Computer Professionals for Peace and Societal Responsibility (FIfF) and co-founder of Bits&Bäume.
The image collection by Lone Thomasky and Bits & Bäume have already been featured across various media outlets and other public forums. As more people use these images, we hope that they can support those advocating for stronger policy and legal frameworks for sustainable digital transformation. While the EU AI Act includes some provisions to assess the environmental impacts of AI (such as for high-risk AI systems), these remain limited and largely dependent on self-regulation, meaning that AI companies have little incentive to improve the sustainability of their AI supply chains.
The Montreal AI Ethics Institute 2025 State of AI Ethics Report Vol 7 is an example of one of Lone Thomasky and Bits & Bäume’s image collection ‘in the wild’. The use of ‘Distorted Lake Trees’ from this collection to represent the state of AI in 2025 is fitting, given the increasing awareness of the impacts of AI on the environment that we have seen this year. Their report aims to capture perspectives from those excluded from conversations about AI, such as voices from Canada, the US, Asia, and Africa – places where the environmental impacts of AI are also most pertinent.
Part II, Chapter 7 of the Report is focussed on the environmental impact of AI, in their contribution Burkhard Mausberg and Shay Kennedy state:
“As environmentalists, we believe that ethical AI must include ecological intelligence. That means embedding sustainability metrics into model development, mandating transparent lifecycle reporting, and aligning national AI strategies with climate goals. Governments are beginning to move in this direction but need to accelerate their oversight and avoid the race for the bottom.” – Burkhard Mausberg and Shay Kennedy (Small Change Fund)
Therefore, beyond improving public understanding, these visuals can play a greater role in sparking debates about the real-world implications of AI development. The technosolutionist narratives of progress and innovation turn a blind eye to impacts on our communities, land, and environments inhabited by animals and humans. It is important that there is a greater understanding of the environmental costs of innovation, only then can we have more meaningful conversations about the role (if any) of AI technologies in our society.
“If you’re building or buying AI in 2026, you inherit the full anatomy: extraction, fabrication, operation, disposal. Before approving any flashy AI project, start by answering these questions: Where is it? Who owns it? What does it consume (hour by hour, basin by basin)? Who carries the residuals?” – Priscila Chaves Martínez
Download Lone Thomasky and Bits & Bäume’s image collection here
