Researchers develop 30,000 AI-induced climate mitigation scenarios

Chinese researchers have developed 30,000+ artificial intelligence-induced climate mitigation scenarios, according to a study published on Nature Climate Change journal's July issue.

The published study covers the potential of deep learning (DL) merged with integrated assessment models (IAMs), offering a base for handling climate mitigation scenario generation.

"IAM based scenarios often face challenges such as modelling biases and large computational burden. Here we develop a DL framework to generate key variables through synthetic mitigation scenarios," the study said.

According to the study DL is a "powerful tool for extracting hidden patterns and intricate interactions from high-dimensional data."

It helps generate future scenarios of how greenhouse gas emissions might be reduced by overcoming IAM limitations, such as a lack of data for certain regions.

Another limitation the research shows is the bias that models can retain due to training on existing biased data.

"One key limitation of this study is that our generative DL framework replicates only a subset of key variables, far fewer than the hundreds of variables required for full-scale IAM scenarios."

Another limitation the research shows is the possible bias models can retain due to training on existing data, which may already have biases.

"Our DL models were trained on the existing AR6 Scenarios Database, which may retain or reinforce those biases."

Although deep learning cannot directly replace traditional models currently, it can offer support and enhancement by creating faster, more diverse, and region-specific scenarios, which can be used to help policymakers and developing countries plan climate action.