Digitalization Gone Astray

Source: Generated by DALL-E

Digitalization in general and artificial intelligence in particular is often proclaimed to be the solution to most, if not all, sustainability challenges. We tend to think that digital is sustainable, because the data are just flying wirelessly through the air.

There are indeed a vast number of highly valuable examples of how digital technologies solve important challenges, such as improved power utilization in production, better fish health in the aquaculture industry, and automated analysis of X-ray images in healthcare. However, the unpleasant reality is that all digital technologies leave significant and highly tangible physical footprints on our environment. The storage, transmission, and processing of data requires processing power. Generative artificial intelligence is particularly power-hungry. Electronic devices are also essential parts of the value chain, e.g. smartphones, wearables, industrial robots, servers, computers, monitors, etc. All these electronic devices come from somewhere, and when we are done with them, they go somewhere. The mining industry and waste management is intrinsically related to our digital transformation.

Let us dig into some more facts and details.

All things digital require physical hardware, such as smartphones, computers, servers, smartwatches, monitors, etc. All electronic devices require precious metals, which are obtained through physical mining. Many of these mines are either located in conflict zones, such as Russia and Congo, staffed by children, or in dire conflict with vulnerable ecosystems, such as the Norwegian plans for deep-sea mining.

All digital applications require electrical power. The explosion of generative artificial intelligence has led to a further brutal increase in the need for data centre capacity. According to the International Energy Agency, the data centre industry is on a trajectory to double the energy consumption from 2% of the total global energy consumption in 2022 to 4% in 2026, mostly due to generative artificial intelligence. That means that the global data centre industry alone is estimated to require as much electricity in 2026 as twice the current total electricity consumption of Germany.

Even if the major cloud providers are working towards "net zero carbon footprint,” the fundamental problem remains: insufficient production of sustainable electrical power. When big techs soak up all available renewable energy and carbon quotas, they are simply shifting the carbon problem to other players with less financial muscles. To meet the steeply increasing needs for electricity, all the major cloud service providers have recently announced heavy investments into nuclear power. Indeed, nuclear power is a key energy resource for the future, but is the development and expansion pushed forward by the right drivers?

Access to clean drinking water is a geopolitical fire torch. In the frenzy to adopt digital technologies, it is easy to forget that the technology industry is a major consumer of clean water, both in the production of electronics and for cooling of data centres. Meta is building a new data centre in Toledo in Spain. This data centre is projected to require 660 million litres of water for cooling per year, in one of Europe’s dryest regions.

Digital applications require devices. All commercially oriented hardware providers in the market have recently launched high-profile marketing campaigns for the latest “AI-powered-something.” When you buy new devices, what do you do with the relatively new and still very usable device that you no longer need? Less than one fifth of consumer electronics is recycled. The rest ends up in landfills, very often on the African continent, such as Agbogbloshie in Ghana. Child labour is not eradicated with digitalization, it just changes format. 18 million children are according to WHO currently at immediate health risk because they work in the informal waste management industry to dismantle smartphones and computers, a highly toxic job.

There are no silver bullets that can solve this imbalance between technological progress and environmental footprints. However, we can all contribute:

i)                        Keep electronic devices for longer. When you do need to replace them, make sure to either hand them off for restoration and a second life or make sure they are dispatched for recycling.

ii)                      Modesty in consumption is good for the planet, also when it comes to digital tools and content. Different generative artificial intelligence models (LLMs) have vastly different environmental impacts. Choose wisely.

iii)                    Efficient code used to be a quality stamp of experienced developers. Build a culture for efficient software engineering in your company, including the AI teams.

iv)                    Include the environmental footprint of your digital transformation efforts into the sustainability reporting to promote transparency and awareness.

The digitalisation toolbox is just that, a toolbox. Artificial intelligence constitutes a key part of this toolbox, but it is by no means sustainable per se. Whether it becomes an environmental villain or a force for good depends on how we choose to apply the toolbox. It is high time to make conscious choices.