With the exponential proliferation of ChatGPT and similar Generative AI tools — and their seamless incorporation into our lives — comes the intensive task of keeping these systems running. Around the world, national governments and large tech companies are planning for future energy needs and building as many data centres as are necessary. Japan is accounting for higher energy consumption due to AI in their national energy plans. In the US, Microsoft is encouraging the reopening of the site of the worst nuclear disaster in US history to power its data centres. Chile’s capital city, Santiago, will see 16 data centres for US tech giants built over the next few years - if their water doesn’t run out by then!

In the face of increasing climate change and catastrophe, the environmental impact of AI technologies has become more prominent and a greater cause for alarm.

Data centres require an incredible amount of electricity to process queries and, indirectly, keep the facilities running. A single ChatGPT query, for example, requires 2.9 watt-hours of electricity, compared with 0.3 watt-hours for a Google search; that adds up over 1 billion queries per day!

It also takes a lot of water to process queries and keep data centre servers from overheating: for GPT-4 to process every 100-word query, it needs 519 millilitres, or just over a bottle, of water; cumulatively, that’s millions of litres of water for cooling and processing! It doesn’t help at all, either, that these centres are usually built in places that already experience “high levels of water stress”.

With data centres already responsible for about 1% of global greenhouse gas emissions, the increasing rise of data centre carbon dioxide emissions could represent a “social cost” of $125-140 billion by 2030. That’s strike three of many. This paints an increasingly dire picture of AI's environmental impact. What we don’t see when we ask ChatGPT to summarise a lengthy report or ask Midjourney to generate an image in the style of a favourite artist is the strain we are putting on our already-depleting environmental resources and the trade-offs that communities living around data centres have to endure. The River’s Ledger is a story that highlights the reality we are hurtling toward with the unfettered development and deployment of Generative AI technology.

We already have proof of the burden that current AI usage is placing on the environment and people living in affected areas or near data centres. In the story, mythic avatars of Indian rivers and water bodies lament the impacts of data centres on the communities they serve: power and water shortages, higher real estate prices, land constraints, encroachment on usable land, and improper disposal of industrial waste. And that’s just the socio-environmental issues! Living near data centres comes with other kinds of ailments, like having to endure noise pollution and health issues like an uptick in blood pressure and hypertension due to stress induced by the centre’s constant hum.With the growing ubiquity of Gen AI, especially in the form of multi-model ‘agents’ that can plan, execute tasks, and collaborate with other Gen AI agents to perform actions on behalf of humans at their convenience (and soon without human oversight), energy and water usage are likely to increase exponentially, along with the other adverse effects of erecting data centres. In The River’s Ledger, the social and physical impact of this world, which operates on Gen AI agents, is reflected materially in the hyper-transactional nature of everyday life, the quality and composition of the rivers, and the subpar quality of life that those less privileged have to endure.

Because they are so steeped in furthering their AI agenda, the tech executives in The River’s Ledger do not, at all, notice the impacts of unfettered innovation and maintenance of their AI systems. They only finally notice when the rivers take revenge in a way that directly impacts their products and services, serving as a reminder that AI development and deployment are at the mercy of planetary boundaries and must therefore respect them. Our earth, and what it offers, is not an infinite resource that can be plundered without care or consideration.

Nila, an activist scientist, is able to understand the river’s intricacies because she places equal importance on ground truth and indigenous practices of engaging with nature and place, as she does on traditional scientific enquiry and data analysis. By observing the rivers’ patterns and decoding them with her grandmother’s notes, she notices how the river is turning on the tech. In their haughtiness, the tech executives dismiss her, despite her findings being just as grounded in rigorous scientific method. 

What unfolds is a hilarious yet tragic series of events that compel the tech executives — and, to some degree, everyday users of Gen AI agents — to confront the fact that the planet’s limits have been reached.

‘The River’s Ledger’ does not advocate for eliminating tech and data centres; it recognises the importance of such innovation and advancement. However, data and AI infrastructures depend on the health of the river; the overexploitation of environmental resources eventually stifles the speed and trajectory of technology development because there are insufficient resources to power them.

The story urges us to reflect on what is feasible, given the finiteness of the Earth’s resources. It also asks that we re-examine the ‘luxury’ of seamlessness and convenience that Gen AI agents will make possible, and perhaps relinquish the hold it has on us to slow down the toll that developing, deploying and running AI is having on our planet.

Reversing, or at least stalling, the environmental impacts is mainly in the hands of these huge tech companies and partly in the individual choices around how much we use AI. Our consumption also drives production, and we must reconcile with that reality and push back against narratives and rhetoric claiming progress will be impeded if we do not all jump on the AI train.

Advancements in technology must respect planetary boundaries. The price of so-called progress cannot be the planet.

The River’s Ledger is a call to:

• Evaluate the trajectory of AI innovation, weighing the environmental impacts of relentless AI development and upkeep. This means taking into consideration the material realities of data and AI infrastructures, particularly their effects on surrounding human communities and species. The universe within The River’s Ledger is powered by agentic AI, a rapidly growing innovation that currently requires and will continue to require immense compute power and resources. To support their agentic AI products and services, the Megh-Kendra tech company keeps their data centre running at full tilt, negatively impacting surrounding communities and depleting resources. A systematic evaluation of the immediate and future impacts of such technologies, along with the necessary infrastructures to support them, would offer a more balanced view of the trade-offs and benefits of developing, deploying, and running such technologies.
  
  An integral part of evaluating the trajectory of AI innovation is to also look beyond abstracted scientific “quantifiable” indicators to assess the impact of Gen AI systems, and include more holistic, indigenous knowledge systems and ways of seeing. These knowledge systems and practices, which may be place-based (relevant to a specific community and/or physical area) can be just as insightful and beneficial for deployment and longevity of the technology. In The River’s Ledger, Nila’s grandmother urges her to observe and pay attention to localised ecological indicators that are more subtle and less likely to be captured in traditional monitoring systems, to understand what is happening with the rivers.
• Recognise that the convenience Gen AI technology affords consumers comes at a cost; critical to this is building consumer awareness around the risks and harms of unmoderated consumption. What is likely to be omitted from traditional monitoring systems, analyses of the long-term benefits of Gen AI technology and related hype narrative are the costs to people and the planet in the long-term, leaving our outlook on eventual progress incomplete. The convenience these technologies afford consumers comes at a price that we are willing to pay in the short term, without considering what continued and perpetual use means for us and the Earth in the long run. While the responsibility lies with tech companies to be more discerning and intentional in the development and deployment of technology, it also lies with the consumers of these technologies to think critically about their usage. When we think of ‘progress’ for the future, as consumers, we also need to assess whether the longer-term benefits are worth the harm we are contributing to now and will continue to contribute to in the future. We should not allow our imaginations to be so easily swayed that rejecting this technology seems regressive and/or impossible.