By now, everyone is aware that the use of AI consumes a lot of energy and therefore does not necessarily have a positive impact on the environment. While some attention is already being paid to this issue, 55% of Dutch people believe that even more should be done. This applies primarily to companies that use artificial intelligence on a large scale but simultaneously claim to support sustainability. Can companies use AI to solve their own sustainability challenges? We’ll explore this in more depth in this article!
What is the environmental impact of AI?
The environmental impact of artificial intelligence is far greater than most people realize. Currently, our digital lives already account for 4% of global emissions. That makes it more polluting than aviation. Research shows that with new AI technology, this percentage could rise to 14%. The use of AI therefore certainly has a major impact on the climate. For example, training ChatGPT in 2024 resulted in 500 tons of greenhouse gases. That is equivalent to a thousand cars driving a thousand kilometers. Every question you ask ChatGPT results in 4 grams of CO2 emissions. That is equivalent to the production of a plastic bag and ten times as much as a regular Google search. If this continues at the same rate, by 2027 ChatGPT will consume as much electricity as the entire Netherlands combined.
Companies that use AI
Most emissions, of course, come from companies that use artificial intelligence on a large scale. Research shows that a large proportion of the Dutch population believes that companies which claim to be sustainable but also make extensive use of AI lack credibility. This is not surprising, since the large-scale use of artificial intelligence is not particularly sustainable at present. In some cases, experts are already advocating for an AI tax, similar to the one currently in place in the aviation industry. After all, it is difficult to place the responsibility solely on companies or solely on consumers. It is therefore important that the government monitors and regulates the use of AI and its impact on sustainability through appropriate measures.
Can we let AI solve its own sustainability problem?
Could we use AI to solve the sustainability problem that AI itself causes? That is the question researchers are currently working hard to answer. Despite the fact that artificial intelligence consumes a lot of energy, it has already solved or mitigated many sustainability issues across various sectors. For example, AI can significantly reduce energy consumption in buildings through the real-time use of heating and cooling. This can already yield savings of up to 40% per building—ideal for large office buildings or factories. Energy consumption is also starting to drop in data centers thanks to AI, which makes processors more efficient. For example, new processors from NVIDIA already use 20 times less energy than regular processors!
Artificial intelligence is also crucial for the deployment of smart energy grids. It does this by balancing supply and demand in real time and better integrating renewable energy sources into the grids. This also helps prevent overloading of the energy grids caused by solar panels, for example. Finally, AI accelerates the development of clean technologies. Think, for example, of improved solar cells and alternative proteins. Over time, these kinds of innovations will lead to significantly lower emissions and pollution!
So what about Generative AI?
The type of artificial intelligence that consumes the most energy is generative AI. Think, for example, of image generation. Tasks like these can sometimes consume as much energy as fully charging a phone. Generating images with AI is therefore not necessarily high on the sustainability scale. Text generation, on the other hand, is 160 times more energy-efficient, so that task isn’t as bad. Still, the AI models themselves can also be a lot more energy-efficient, according to research by DeepSeek. Data centers used for AI processing account for 2.5% to 3.7% of global emissions. Currently, AI’s share of the data technology sector’s global emissions is still relatively low, at 0.1%. That is comparable to Sweden’s annual emissions.
Is an AI tax a good idea?
Many people are calling for an AI tax, but is this actually a good idea? We now know that artificial intelligence can actually help save energy and reduce emissions, especially for large companies. But will large companies still use these applications in the same way if an AI tax is introduced? After all, this would cost them a lot of extra money. This could therefore also hinder innovation in the field of sustainable technologies. Finding a good middle ground here is important, rather than imposing a generic AI tax. Governments should specifically examine AI-related energy consumption per company and the savings these innovations generate. This way, we can all ensure that AI contributes to sustainability within companies!