"Artificial intelligence can boost economic growth in an unprecedented way, fuelling the environmental crises."

The article reflects the personal opinions of the authors and does not necessarily reflect the position of SCNAT.

Artificial Intelligence (AI) is on everyone’s lips. While some organisations have started to ethically question some of AI’s developments, most remain confident that the technology holds great potential to tackle key issues of our time, including the environmental crises. Indeed, various applications of AI are considered promising for fighting climate change, such as machine learning (ML) applications to improve buildings’ energy efficiency or forecasting systems for electricity supply and demand. However, when one dives into the real and potential environmental implications of AI, the picture is not all that bright. Indeed, AI not only has considerable direct environmental impacts, but also could play a major role in boosting and sustaining economic growth, thus strengthening the root of the environmental crises.

AI relies on the same hardware and infrastructure as other digital technologies (e.g. computers, servers). This means that it also shares their environmentally destructive consequences, for instance those relating to the extraction of rare earth elements, the energy-intensive production processes, and the massive e-waste stream. AI also has its unique environmental costs, however. ML models are based on learning from data, some from a lot of data. This results in long computing times and large storage demands, in turn leading to high energy consumption. This also drives the proliferation of data centers, with the corresponding high water and land usage.

There is no decoupling of economic growth and environmental pressures

However, this is only one side of the story. AI can also play an environmentally destructive role by accelerating economic growth. Growth was and still is a key driver of the environmental crises, as there is yet no prospect of a sufficient absolute decoupling of economic growth and environmental pressures. While technology has always been an important driver of economic growth, AI could turn out to uniquely and more strongly accelerate economic growth compared with other technologies. One first mechanism through which it could do so is by accelerating the generation of ideas and new knowledge. The fashion company Shein, for example, uses AI to monitor social media in real time, detect fashion trends, and generate design suggestions for new clothing items. This use of AI to turbocharge the creation process allows Shein to release over a thousand new items per day on their website; their revenues increased by 250% between 2019 and 2020 (1). Another mechanism is the self-learning characteristic of AI. Some ML models are programmed to continuously learn, becoming increasingly accurate the more data they are fed. This means that, unlike other kinds of capital that depreciate over time, ML can perpetually increase its contribution to productivity, thereby continously and increasingly driving economic growth.

The growth-driving potential of AI is strongly emphasised by large consultancy firms and governments , and their assumptions lead them and others to sustain high investment levels in the technology. Pricewaterhouse Coopers (PwC), one of the leading international accounting firms, estimates, for example, that the use of AI will lead to an increase of 14% of the global GDP by 2050 (2), and the European Union describes AI as a strong force for productivity in numerous of its communications.

Avoid environmentally destructive lock-ins

There is an obvious contradiction in the discourse surrounding AI: on the one hand, AI is presented as a useful technology to alleviate climate change and other environmental crises, while on the other hand, it generates destructive environmental impacts – also by increasing economic growth. If this contradiction is not addressed very soon, investments made in AI today might create lock-ins in a set of infrastructures that are environmentally destructive and growth pursuing (3). This could take us even further away from creating the sustainable society we need.

Note: This article is based on the research of Marion Meyers “A Degrowth Perspective on Artificial Intelligence - Analyzing the appropriateness of Machine Learning to a Degrowth Context” and her paper “Artificial intelligence in a degrowth context. A conviviality perspective on machine learning” GAIA 33/1 (2024): 186–192.

-

Marion Meyers is an MSc graduate in Science, Technology, and Policy from ETH Zürich. The economist Irmi Seidl is the Head of the “Economics and Social Sciences” Research Unit at the Swiss Federal Institute for Forest, Snow and Landscape Research (WSL).

References:

1. Bloomberg News (14. Juni 2021). How Trump’s Trade War Built Shein, China’s First Global Fashion Giant.
2. Anand, R.; Gerard, V.; Euan, C. (2017). Sizing the Prize – What’s the real value of AI for your business and how can you capitalise? PwC Report
3. Robbins, S.; van Wynsberghe, A. (2022). Our New Artificial Intelligence Infrastructure: Becoming Locked into an Unsustainable Future. Sustainability 2022, 14, 4829.