AI will be an eco-killer

High hopes are pinned on AI and digitalization. They are expected to solve not only all economic and other problems, but also those of environmental protection. This is often still too inconvenient or unfamiliar to us all, and now seemingly miraculous technological solutions are on the horizon. These solutions are to be made possible by making resource use and energy consumption more efficient, by optimizing logistics along the entire value chain, right down to waste management, or by making houses, industrial production, and even fertilization and pesticide use smarter. If AI can simplify production, it can, for example, save a great deal of material, reduce transportation, and thus also energy. This is intended to enable resource-efficient growth that reconciles economic and ecological concerns.

But the opportunities AI offers for environmental protection are similar to those for democracy and the economy: they are accompanied by risks. Currently, AI is more likely to become an environmental killer. This is because ever-newer applications and services initially require large amounts of materials, energy, and often space. The enormous data centers being built to train AI with vast amounts of data are particularly energy-intensive. In a world of Trumps and Putins that is increasingly relying on fossil fuels again, this means not only more material consumption but also even more fossil fuel consumption.

The problems with climate, biodiversity, and almost all other environmental challenges demand a timely and complete phase-out of fossil fuels, not only for electricity but also for heating and transportation, agriculture, and cement and plastics production. Digitalization and AI must also meet the minimum ecological requirement of being powered by renewable energy. And this must happen soon – climate change and biodiversity loss don't give us until 2050, as many believe. Because renewable energy is not infinitely available, digitalization also requires mechanisms that encourage efficient and frugal energy use. This is pretty much the opposite of what AI has led to so far, as it invites numerous gimmicks, precisely because this offers a huge business opportunity and widespread entertainment.

However, excessive electricity consumption is not only problematic because of fossil fuels, but also because the increasing electrification of all areas of life makes us more vulnerable as a society – after all, a power grid can be easily cut off by terrorists or hostile states. Sustainable digitalization therefore requires incentives to move away from fossil fuels and towards the responsible use of energy. This also means tightening the EU Emissions Trading System, rather than weakening it, as the European Commission has just proposed. Emissions trading is the main instrument in the EU for gradually phasing out fossil fuel energy production and encouraging businesses and individuals to use energy efficiently. This also promises to prevent increasing dependence on the US, specifically on US liquefied natural gas, not only in terms of data but also in terms of energy.

AI also has the potential to enable farmers to use fertilizers and pesticides more precisely. However, if this simply leads to cost savings and the cultivation of more land, the ecological benefit is lost. Climate and biodiversity protection demand that nature be given more, not less, space so that biodiversity can recover and, for example, peatlands can once again store more greenhouse gases. To achieve this, not only must the energy supply for agricultural AI become post-fossil, but the path to ecological agriculture must also be pursued consistently, instead of being slowed down as has been the case recently. Above all, livestock farming and thus feed production, which accounts for the largest share of agricultural land use, must be drastically reduced. One way to achieve this would be a separate emissions trading system for livestock farming at the EU level.

Another environmental problem posed by AI remains unresolved. While AI and digitalization possess the much-cited potential to promote knowledge and discourse, they also threaten to undermine human attention spans and exploit and encourage emotionally charged reactions, particularly on social media. Digitalization, in its various applications, often thrives on amplifying emotional responses and prioritizing unlimited consumer sovereignty. This not only jeopardizes the functioning of a liberal democracy based on rationality and compromise, but also environmental protection. Solving highly complex, long-term problems like climate change and biodiversity loss depends on dealing with facts and a willingness to engage in critical, nuanced thinking.

Unfortunately, these are already scarce resources for us humans. As behavioral science has repeatedly demonstrated, factual knowledge, rational thinking, and values play only a limited role in human motivation. Self-interest and emotions such as convenience, habit, denial, or the tendency to make excuses often override them. Therefore, neither sustainability itself nor an ecological digitalization will happen automatically. If we want them, we have to fight for them. And we must strengthen all the mechanisms that a fact-based internet requires, also for the sake of environmental protection.

This article is also published on Die Zeit. illuminem Voices is a democratic space presenting the opinions of leading Sustainability Thought Leaders, their views do not necessarily represent those of illuminem.

The world needs sustainability knowledge. At illuminem, no interest group or shareholder can influence our work. Thank you for supporting our mission to make high-quality and independent sustainability information free for all. Every contribution helps. Thank you for donating today.