The rapid expansion of AI technology has put an immense pressure on natural resources, from processors to electricity and water, to power the computer data centers that drive the tech. In fact, a report by the MIT Sloan Management Review shows that if current trends continue, AI would soon become one of the largest contributors to carbon emissions. With this in mind, it is no surprise that training AI models requires vast amounts of computing power, with OpenAI’s GPT-3 model estimated to generate 552 tons in carbon emissions- the equivalent of the yearly emissions of 120 U.S. cars. And the numbers get more concerning: a single #ChatGPT query can generate 100 times more carbon than a regular Google search!
Despite the resource-hungry nature of AI, a recent research paper by the University of California, Irvine, MIT, the University of Kansas School of Law, and others, has shown that AI could in fact be much more sustainable than human labor. Titled “The Carbon Emissions of Writing and Illustrating Are Lower for AI than for Humans”, the paper compared the average carbon emissions of a human writer or illustrator to the emissions from AI models performing the same tasks. The researchers discovered that AI models BLOOM and ChatGPT (GPT-3) were 1,500 and 1,100 times less impactful than a US resident per page of text produced. Similarly, AI models DALL-E 2 and Midjourney emit approximately 2,500 and 2,900 times less CO2 than a US resident illustrator.
So, it appears that using AI to perform tasks may be far more beneficial to the environment than having humans do them. However, the strain AI places on natural resources still needs to be managed better, and the MIT Sloan Management Review report highlights some innovative ways that AI companies are doing this. As AI models continue to become more powerful and efficient, it is likely that they will be able to solve a lot more environmental problems than they cause.
