Will AI Save the Climate or Destroy It? The Big Problem with AI

Why does artificial intelligence increase carbon dioxide emissions?

The problem is that AI tools like ChatGPT, Gemini, Midjourney, and others require a lot of processing power. To ensure that, tech companies are investing billions in giant data centers — buildings filled with tens of thousands of computers performing calculations. These servers require massive amounts of electricity — both to power the computers themselves and to cool them — which can emit significant amounts of carbon dioxide, depending on the energy source used to power them. Building data centers also involves emissions Production and transportation of equipment necessary for its operation.

According to the International Energy Agency, total electricity consumption in data centres could double by 2026, reaching a level equivalent to the total energy consumption of Japan. Research firm SemiAnalytic estimates that by 2030, data centres will consume 4.5% of the world’s electricity. Water use is also significant – one study estimates that AI could be responsible for using up to 6.6 billion cubic metres of water by 2027, nearly two-thirds of England’s annual consumption.

A report by the Climate Action Against Misinformation coalition of environmental groups includes a statement saying that “people will be outraged when they see how much energy AI will consume in the coming years, and how the region will be flooded with misinformation about climate change.”

Tech giants are trying to invest in renewable energy.

However, the environmental costs of AI can be reduced by using clean energy to power data centers. That is, either renewables or nuclear power. Amazon, which — in addition to being the world’s largest retailer — is one of the world’s largest cloud providers, is also the world’s largest buyer of renewable energy. However, there is a problem: AI’s energy consumption is growing at a rate that often makes it necessary to use old, dirty energy sources, such as coal-fired power plants, because there are simply too few clean sources.

By the end of this decade, global renewable energy resources are expected to triple to meet governments’ climate targets. But achieving that target is now in question. The International Energy Agency, the world’s energy regulator, has warned that while global renewable energy capacity is set to grow at its fastest rate ever in 2023, current plans will only be enough to double capacity by 2030.

Onshore renewable energy projects, such as wind and solar farms, are built relatively quickly – they can take as little as six months to develop. But the slow pace of permitting, coupled with the weaknesses of ageing power grids that may not be able to cope with connections to many dispersed sources, can slow the process down significantly. This has raised concerns about whether eRenewable energy can keep up. Developing artificial intelligence. According to the Wall Street Journal, the largest technology companies already use a third of the energy generated by U.S. nuclear power plants.

Will AI become less voracious?

However, rapid advances in artificial intelligence – both in terms of the algorithms themselves and the computers they are trained on – offer hope that the rapid increase in emissions can be halted. New models require (relatively) less computing power to train, which means they require less energy than their predecessors.

But for now, this doesn’t translate into a reduction in electricity consumption. Instead, the new AI models are more efficient, using roughly the same amount of energy as their predecessors. In economics, this phenomenon is called the “Jevons Paradox,” after the economist who observed that James Watt’s improvement of the steam engine, which allowed much less coal to be used in industrial production, led to a huge increase in demand for this fuel, because when the price of steam power fell rapidly, new uses were discovered that were not profitable when energy was expensive.

Certainly, rather than being a tool for mitigating the effects of climate change, It introduced a huge new element of uncertainty. For all climate models. Our success in stopping the planet from warming depends largely on how AI evolves in the future.