More than 60,000 studies were crunched using machine learning by researchers in Germany. The findings are not comforting.
Climate change is going to get worse if we don’t act soon. While this is not breaking news, it was the emphasis of a, released in August. The IPCC report warns that rising temperatures will affect every region of the planet.
A new paper, published October 11 in the journal Nature Climate Change, adds some specificity to that forecast. Using machine learning techniques to analyze more than 60,000 climate change-related studies, researchers in Germany estimate that 85% of the population is affected by human-induced climate change. The study was led by Max Callaghan of Berlin’s Mercator Research Institute on Global Commons and Climate Change.
“There is overwhelming evidence that the effects of climate change are already being observed in human and natural systems,” the paper reads. “We estimate that responsible anthropogenic impacts may occur in up to 80% of the world’s land area, where 85% of the population resides.”
study goes onThe United Nations Climate Change Conference in Glasgow, which runs from 31 October to 12 November. COP26 will bring together world leaders including US President Joe Biden and UK Prime Minister Boris Johnson, but not China’s Xi Jinping in particular, in the hope that they will make new commitments to reduce carbon emissions. The Paris Agreements were reached at COP21 in 2015, and observers expect more ambitious commitments to carbon neutrality can be agreed upon in Glasgow.
Machine learning is a type of artificial intelligence that gets smarter as more information: Think speech-to-text software, which can more accurately listen to more bells and whistles. Callaghan and team aim to uncover not only the planet’s plight as the effects of climate change become more known, but also to use machine learning to reveal gaps in scientific research.
The researchers fed 2,373 abstracts on climate change-related papers to machine learning software called BERT (or “bidirectional encoder representations from transformers”). After digesting information on climate change, the algorithm identified studies that could show the effects of climate change, even if those studies did not attribute their findings to climate change. The paper referenced one such study on the relationship between the timing of snowfall and the population growth of mammals.
“Our aim is to map all potentially relevant studies on climate-related changes to a list, rather than a list of studies where the relationship between an observed climate trend and specific effects is demonstrated with high confidence,” the paper reads. “While conventional assessments may offer relatively accurate but incomplete pictures of the evidence, our machine-learning-assisted approach produces a detailed preliminary but quantitatively uncertain map.”