经济新闻
独特的AI模型追踪全球碳排放 2026-04-09
Chinese researchers have unveiled a first-of-its-kind artificial intelligence model designed to track carbon emissions across global production, consumption, and natural resources, a move experts say could shift the balance of power in international climate negotiations.
The model was released on Wednesday by the Shanghai Advanced Research Institute of the Chinese Academy of Sciences.
It arrives as China strives to empower domestic companies to meet green energy goals while strengthening its technical standing in global climate governance.
The new tool is a "large language model" — a type of AI trained on massive amounts of data — featuring 32 billion parameters.
In the world of AI, parameters are essentially the "synapses" that enable the model to recognize complex patterns and proceed to make predictions.
The system utilizes five "intelligent agents", which are specialized sub-programs designed for specific tasks.
These include digital simulation to find the most efficient ways to run factories, trade carbon transfer tracking how carbon "moves" between countries, and life cycle assessment to calculate the environmental impact of a product from raw materials to disposal.
It also features agents for natural source accounting to measure carbon absorbed by forests and uncertainty analysis to ensure the data is reliable.
Gao Yunhu, a researcher at SARI, said the AI acts as a "carbon accounting butler". Unlike traditional methods, which are often slow and expensive, this model allows companies to simulate production in real time to find the cheapest ways to cut emissions.
According to Zhang Xian, director of the Division of Global Environment in the Administrative Center for China's Agenda 21, traditional carbon accounting is time-consuming and costly. It also makes it difficult for companies to accurately measure emissions at each stage of production.
He said the new tool, which functions like a corporate "carbon accounting butler", can assess the full life cycle of a product starting from raw materials. Life cycle assessment refers to measuring environmental impacts from production through use and disposal.
Zhang said this could transform emissions reduction efforts from a cost burden into a source of competitiveness by enabling targeted technology deployment.
"By quantifying companies' behaviors and coordinating data along industrial chains and supply chains, the new model can improve market monitoring, quota verification, and policy assessment. It has provided reliable technical support for global green trade and carbon pricing," said Lai Xiaoming, chairman of the Shanghai Environment and Energy Exchange.
The timing is critical for Chinese exporters facing new environmental taxes in Europe.
The European Union's Carbon Border Adjustment Mechanism recently introduced rules that place a price on the carbon emitted during the production of goods such as steel and cement entering the bloc.
Mi Zhifu, a professor of climate change economics at University College London, noted that the EU currently uses "default values" to estimate emissions if a company cannot provide its own data. Because these default values for Chinese products — particularly steel — are often far higher than their actual emissions, Chinese goods can appear less environmentally friendly than they truly are. By providing verified, precise data, the new AI model helps Chinese firms avoid these higher estimated taxes, especially in sectors such as steel, cement, hydrogen, electricity and fertilizers.
A major feature of the SARI model is its focus on consumption-based accounting.
Current international standards usually follow a territorial principle, meaning emissions are blamed on the country where the factory is located.
SARI's model tracks "embedded carbon" — the carbon footprint hidden inside a finished product — to show that the person or country buying and using the product also shares responsibility.
Wei Wei, vice-president at SARI, pointed to China's green exports as a primary example. In 2024, Chinese wind turbines and solar panels created about 2 million metric tons of carbon during their manufacture. However, those same products will help the world reduce carbon by 350 million tons during their operation.
Zhang said traditional accounting attributes a large share of export-related emissions to China while obscuring the responsibility of developed countries on the consumption side. By quantifying these "carbon leaks" in global trade, Chinese officials believe the model provides a more scientific basis for future climate negotiations.
Source: China Daily
