Researchers at the University of Washington have built an AI system that can calculate the carbon footprint of digital devices like smartphones in under a minute. The tool uses two AI agents to scan public data such as product descriptions and life cycle assessment databases. It aims to make carbon footprint information as easy to find as a nutrition label, helping consumers and companies make more sustainable choices. The AI has an error rate of 5 to 19 percent, comparable to human assessments, but it is much faster and more accurate for materials not in existing databases. The researchers note that running the tool uses about as much energy as brewing a cup of tea, addressing concerns about AI's own carbon cost. This could be a step toward standardizing carbon labels for electronics, which would give buyers a clearer picture of the environmental impact of their devices from raw material extraction to disposal.
