Artificial intelligence can help curb climate change by optimizing energy production, transport, industry, heating, even food production in such a way that climate emissions are produced as little as possible.
However, artificial intelligence itself is a guzzler of energy, water and raw materials, so in the worst case it can take you from the ditch to the source. The harmfulness or blessing of artificial intelligence depends on whether it can be used to reduce environmental harm in other fields more than artificial intelligence causes.
We asked the professor of strategy and innovation about it SpongeBob from Ritala from LUT University of Business and Economics and from an expert at Sitra From Lotta Toivosen.
Different assessments of climate effects
Ritala and Toivonen state that there are different estimates of artificial intelligence’s climate effects and so far it is not possible to reliably assess the overall benefit or harm.
But the fact is that of large language models and generative artificial intelligence energy consumption is high, and the use of artificial intelligence is increasing rapidly.
According to one estimate, the share of data centers in 2026 would be about 6 percent of US electricity consumption, Ritala says.
For example, the energy consumption and greenhouse emissions of Google and Microsoft have increased enormously with the construction of data centers required by artificial intelligence, writes the US media company NPR.
Much depends on whether clean energy is produced for the data centers. For example, in Finland Google’s data centers run almost entirely on renewable energy.
– Somewhere in the world there are data centers that run mainly on fossil energy. The fact that it won’t come to Finland is no consolation, Ritala says.
Even bigger emissions than air traffic
It is difficult to separate the climate emissions of artificial intelligence from other emissions caused by digital use. For example, only part of the software can be artificial intelligence.
International Panel on Climate Change estimate that the data centers and communication networks required by the internet cause an estimated 1 percent of the greenhouse gas emissions related to energy use. Their use and the use of digital devices account for 6–12 percent of energy consumption.
The billions of devices connected to the Internet may be responsible for up to 3.5 percent of greenhouse emissions by next year, the researcher estimates Hannah Ritchie In Our World in Data.
For example air transport climate emissions are about 2.5 percent of all emissions.
Does efficiency increase production?
Different in studies it is estimated that about 15 percent of the world’s greenhouse gas emissions could be reduced with the help of artificial intelligence. Overall, the goal is to halve emissions by 2030.
This is possible because with the help of artificial intelligence, data can be collected and processed more extensively and efficiently than before. That in turn help use resources more efficiently than at present.
Artificial intelligence is also useful in climate research. Climate models have a lot of data to process, and artificial intelligence increases computing capacity. Artificial intelligence can also be used to simulate the effects of climate change.
However, the net benefit to the climate from artificial intelligence depends on what the artificial intelligence is used for, Ritala points out.
The effect is positive when artificial intelligence is used to optimize energy networks and production and improve the efficiency of industrial systems and processes.
– But the matter is more complicated. When we do things more efficiently, we are often able to produce more.
In the end, the effect depends on whether we reduce total emissions or what we do, Ritala states.
Not all use of AI is optimization
A large part of the use of artificial intelligence is not optimization of production that improves energy efficiency.
Artificial intelligence used in many home appliancesin refrigerators, toothbrushes, lighting. For example, a smart refrigerator registers the contents of the refrigerator, a smart toothbrush can be used to ensure that the teeth are properly brushed, and artificial intelligence optimizes the lighting.
You can also entertain yourself, for example, by chatting with ChatGPT or creating artificial intelligence images, which consumes a lot of energy.
– Various artificial intelligence tools are integrated into all kinds of consumer applications, Ritala states.
Toivonen emphasizes that when devices or services are invented, the starting point should be to openly talk about both risks and benefits. In addition to environmental risks, artificial intelligence also involves, for example, data protection risks.
There are other concerns than climate emissions
The only environmental problem of artificial intelligence and digitization is not the skyrocketing energy consumption. For example, the development and maintenance of language models require a lot of waterRitala states. Water is used, among other things, to cool servers in data centers.
This is also related to climate change to the extent that increase in drought is one of the consequences of climate change.
And when everything becomes digital, more batteries and devices are needed, which means that the consumption of raw materials increases. Data centers with their equipment also consume raw materials. The same is true if production increases as efficiency increases.
– If it is predicted that the energy consumption of artificial intelligence will increase tenfold in a few years, what about from then on? Will it increase tenfold again? Ritala asks.
The biggest responsibility lies with service providers and legislators, but even a single person can make a difference, Toivonen says:
– You can carefully buy long-lasting, repairable and recyclable devices and watch videos in moderation, because video images are energy-intensive. Prefer fixed broadband instead of mobile data. Close unnecessary digital devices and data connections.
He also needs a discussion about how much raw materials are needed for digital devices and how much waste is produced.
For example, 1.4 billion smartphones are sold worldwide each year. About 80 different raw materials are needed for them, almost half of which are metals.
– And we only use a smartphone for 2-3 years on average.
Toivonen points out that the use of natural resources, climate change and loss of nature go hand in hand. About 40 percent of the world’s electricity is produced by renewable sources, but it would not be problem-free even if all energy were renewable.
– It is not enough that we produce without emissions, if at the same time, for example, we cause destruction of nature when we take raw materials from nature.
However, Toivonen is optimistic and believes that we are smart enough to use artificial intelligence wisely.
– It requires that issues be discussed and that we dare to bring up different perspectives, he states.