๐๐จ๐๐ฌ ๐๐ ๐ซ๐๐๐ฎ๐๐ ๐๐ง๐๐ซ๐ ๐ฒ ๐๐จ๐ง๐ฌ๐ฎ๐ฆ๐ฉ๐ญ๐ข๐จ๐ง ๐จ๐ซ ๐ข๐ฌ ๐ข๐ญ ๐ฉ๐๐ซ๐ญ ๐จ๐ ๐ญ๐ก๐ ๐ฉ๐ซ๐จ๐๐ฅ๐๐ฆ?
Analysing German regions between 2012 and 2023, the study combines a novel approach โ web scraping from CommonCrawl and a supervised NLP model for identifying AI and sustainability adoption from company websites โ with regional data on energy consumption.
Our findings offer an intriguing perspective: regions with higher AI adoption show significantly lower levels of industrial energy consumption. However, this effect has weakened over time, with stronger energy-saving effects observed during 2012โ2017 than during 2018โ2023. Importantly, the effects are not uniform across regions. Energy savings are more pronounced where both large and small firms adopt AI, highlighting the importance of broad-based technological diffusion. The energy-reducing impact of AI also becomes stronger at higher levels of relatedness to green technologies.