Climate and computing power in education – how AI can both burden and enrich our planet
More and more schools are embracing artificial intelligence (AI). Think of smart learning systems, adaptive testing and personalised learning pathways. But what impact does that actually have on the environment? The report “Klimaat en rekenkracht” (Climate and computing power), a study by consultancy Berenschot commissioned by Kennisnet, sets out the possible environmental effects of AI in primary and secondary education. What does it show? AI uses more energy and water than is often assumed, and that comes with considerable CO₂ emissions. This blog discusses the main findings and how you can deal with them as a school.
From computing power to CO₂ emissions
AI requires a lot of computing power. All those calculations – think of training models or processing data in real time in the classroom – need electricity. According to the report, the environmental impact of AI in a hypothetical scenario in which AI is used extensively for an entire school year is comparable to:
- The annual water consumption of 1,440 households
- The electricity consumption of 6,829 households
- CO₂ emissions equivalent to 2,566 petrol cars, each driving an average distance
These are large numbers that show that AI not only has the potential to improve our education, but also comes at an environmental price.
Where do these figures come from?
Commissioned by Kennisnet, Berenschot researched various forms of AI use in primary and secondary education. In their calculation models, they assumed that AI is used for a wide range of applications, from personalising learning materials to supporting teachers with administrative tasks.
A series of assumptions was made about, among other things, the duration of AI use, the type of hardware and the number of pupils. Of course, those assumptions are always open to debate: not every school uses AI equally intensively, and not every AI system runs at full power all day.
Why this report matters for schools
The report does not address the question of whether or not AI should be introduced in the classroom. The study is mainly intended to raise awareness and to support educational institutions in making well-considered choices.
Some possible pointers:
- Use AI in a targeted way
Not every educational problem needs to be solved with AI. By focusing on the areas where AI really makes a difference, you avoid unnecessary energy consumption. - Choose energy-efficient hardware
Server rooms and data centres are often out of sight of pupils and teachers. Yet the efficiency of that hardware has an enormous influence on total energy consumption. - Be smart with data
An AI model doesn't always need to be trained on gigantic datasets. Being mindful of the amount of data can considerably reduce the environmental impact. - Sustainable partners and suppliers
Look at the climate goals of the parties you work with. With many cloud providers, for example, you can choose data centres that run on green electricity.
Learning and becoming more sustainable
The use of AI in education is promising. It can challenge pupils at an individual level, relieve teachers and enable new forms of education. At the same time, the report shows that we cannot view this development separately from its environmental impact.
Still, the beauty of technology is that it keeps developing – often faster than we think. A great deal of work is being done, for example, on models and data centres that use less energy. And the school board can play an important role by paying attention to sustainability performance when selecting AI applications or ICT suppliers.
Conclusion
AI offers countless opportunities to innovate education and improve the quality of learning. But as with any technological advance, it is wise to shed light on the downside too. With the insights from Kennisnet and Berenschot, schools can now better weigh up when and how AI contributes to their educational goals – and to a sustainable future.
Source:
