An AI-supported system has detected thousands of air pollution incidents in Hangzhou, with authorities claiming it has tripled the speed of pollution source analysis, CCTV reported on 12 March.
Mobile monitoring devices installed on buses and taxis collect real-time air quality data, which AI analyses to generate early warnings, the report explained.
The system can identify high concentrations of PM10 – particles less than 10 microns wide – and their potential sources, then notify relevant personnel to take action.
In Chongqing, garbage transfer stations are also using AI to analyse surveillance footage and detect whether garbage trucks are clean and leak-free before entering the station. However, the station staff told Chongqing Daily they still need to carry out manual reviews to prevent errors. The AI system can also detect when large items like furniture are mistakenly disposed of as regular rubbish, issuing immediate alerts.
While People’s Daily praises AI for improving environmental regulation, its adoption comes with challenges. China Environment News highlights the high initial investment and operating costs, and the time it takes to build and deploy AI infrastructure.
There are also concerns over energy consumption. AI typically relies on data centres, which account for 2-4% of total electricity consumption in China and other major economies, according to the International Energy Agency. Although AI may not use as much power as electric vehicles or air conditioners, its expansion could still strain local power grids.
Moreover, AI-driven automation raises concerns for workers. While governments and organisations believe automation boosts productivity, it can also increase hidden workloads for frontline staff, warns the World Economic Forum. Developing and maintaining digital infrastructure requires behind-the-scenes labour that can add pressure to technicians.
Read Dialogue Earth’s 2023 analysis on the need for citizen help to make China’s emissions-monitoring system effective against industrial pollution.
