Artificial intelligence and sustainability: is AI a threat to the environment?

The Artificial Intelligence Act came into force on August 1st, 2024, promoteing the responsible development and deployment of Artificial Intelligence (AI) in the European Union. AI is increasingly being used in areas such as finance, healthcare, transportation, translation, and is helping to make work more productive in the creation of text or useful templates, emails, and documents, visualisations, advertising, and etc. As AI is increasingly integrated into everyday work processes, it is worth considering the impact AI tools have not only in terms of productivity, but also on the environment.

Is AI a sustainable choice?

One of the rapidly evolving aspects in accountability in environmental protection are sustainability reports. There is also an increaced demand for their accuracy. To complete these reports, it is important to accurately assess the environmental impacts of all business areas, including the use of AI tools. We already know that the use of AI has many benefits in terms of reducing human error, data processing, increased efficiency and productivity at work. These tools help to optimise internal processes, forecast demand and regulate production in real time. On the one hand, the use of AI thus reduces energy and human resource costs for companies thanks to its sophistication. But on the other hand, Google, in its latest sustainability report, claims that 13% more CO2 emissions in 2022 than in 2021 were attributable to the growing development of AI. As in all digital fields, AI requires computing power from thousands of data centres to operate, and therefore consumes large amounts of electricity to meet this demand. The International Energy Agency (IEA) estimates that by 2022, data centres will consume up to 1.3% of the world’s electricity. This may seem like a large figure, but the IEA estimates that the use of data centres has only grown slightly over the last 15 years. Despite much higher demand, data centre efficiency has also improved, partly thanks to the replacement of small data centres by large ones (i.e. the growth of companies such as Amazon, Microsoft, Google and Meta). So while Google, for example, reports that their emissions have increased by 48% over the last 5 years, global data centre energy consumption still remains a small fraction of the world’s total energy consumption. Researchers also estimate that training the OpenAI GPT-3 language model to use OpenAI’s GPT-3 language model has emitted around 500 tonnes of CO2, which is equivalent to charging 60 million smartphones. Again, frightening figures, until you consider that there are currently around 8.5 billion phones in the world, not all of which are smart, not all of which are charged every day, but training the OpenAI GPT-3 model has cost the world the energetic equivalent of charging 0.7% of the total number of phones in existence once.

Future prospects

Given the scale of our energy consumption, the queries we create in ChatGPT or other AI tools are no more harmful to the environment than our searches in Google. However, with the increased use of AI solutions, it is worth disclosing these tools in your annual sustainability report. While the energy consumption of AI may not seem like a significant part of the report, as the technology develops and expands, it is safe to stay one step ahead and be transparent about the use of AI tools. So don’t be afraid to use AI tools in your operations and don’t forget to disclose their use in your sustainability report this year.