AI shows global impacts of climate change

FFor climate science, AI and big data are now part of the research toolkit. A research network led by the Berlin Mercator Research Institute on the global commons and climate change and the company Climate Analytics has entered a new field. Thanks to a state-of-the-art deep learning approach, Max Callaghan and colleagues have evaluated, classified, and mapped more than 100,000 published studies addressing the effects of Climate change Deal. Regardless of whether it is geophysics, biology, agricultural research, or migration and conflict research – the scientific evidence documented across all disciplines of accelerated man-made climate change is now very impressive. The most important result in the trade magazine The nature of climate change Attribution study published using artificial intelligenceAt least 80 percent of the world’s land area, on which 85 percent of the world’s population lives, is now clearly affected by man-made climate change.

Specialized knowledge about the effects of climate change is brought together by individual disciplines – but it has often been loose and incoherent to date. An algorithm programmed by Callaghan and colleagues to map changes to climate change uses these independently found numeric values ​​and aggregates them into a large overall picture. “It would have been impossible to examine these 100,000 prints by hand,” says Carl-Friedrich Schleussner, one of the study’s authors. where The first status report of the Intergovernmental Panel on Climate Change (IPCC) from 1990 The number of articles published annually that directly or indirectly refer to climate change has increased a hundredfold. The report only started a few weeks ago “Working Group I” Sixth Reporting Session of the Independent Expert Committee. In order to publish the current state of research on the physical basis of climate change, scientists at the Intergovernmental Panel on Climate Change needed nearly 4,000 pages—they evaluated about 14,000 publications.

The cloud of knowledge of the global climate is constantly growing, and it is becoming more and more difficult to keep track of things. The researchers’ goal has been to continue to make ever-increasing evidence accessible and socially used, for example for global political processes such as Stream Climate Summit from Glasgow As well as to assess regional risks and coordinate local adaptation measures.

Digital World Map of Climate Impacts

“So we thought about it: How do we get the big literature and big data together?” As Max Callahan says. He and his colleagues found what they were looking for in a BERT speech recognition algorithm (Bidirectional coding representations of transformers) from Google. BERT is an artificial intelligence that can do two things: read and understand. Not only does the algorithm keep track of the meaning of one word, it links it to all the other words in the sentence.