AI training pollutes a lot more than previously thought

MIT took a look at the pollution generated by AI and concludes that deep learning has a terrible carbon footprint. In a new article, Karen Hao explains the training of a single, large AI model can result in as much carbon dioxide emissions as five cars in their lifetimes. To be more specific, a Transformer neural net with 213 million parameters with neural architecture search can emit over 626,000 pounds of CO2 equivalent.

It’s a jarring quantification of something AI researchers have suspected for a long time. “While probably many of us have thought of this in an abstract, vague level, the figures really show the magnitude of the problem,” says Carlos Gómez-Rodríguez, a computer scientist at the University of A Coruña in Spain, who was not involved in the research. “Neither I nor other researchers I’ve discussed them with thought the environmental impact was that substantial.”