New Research Finds Major Inequalities in Climate Science Studies

The study used artificial intelligence to determine that climate change already is impacting 85% of the world’s population.

Clark Dry Lake, Anza Borrego Desert State Park California, USA
Alan Majchrowicz / Getty Images

“I’m only human.” Everyone has probably uttered these words at one point or another. And for good reason: Human beings are flawed. They get tired, bored, hungry, and weary. In other words, they have limits. And when they reach them, that’s it. Game over.

That’s why many scientists are using computers to conduct their research, including an international team of researchers who recently set out to quantify the impact of climate change on the world’s population. To do so, they would have to comb through hundreds of thousands of studies on climate change in order to identify, classify, and map climate impacts across the globe. "Big literature," the scholarly equivalent of big data, is the ballooning collection of scientific literature within many fields. Sorting through them has become an impossible task for even the most dedicated scientists.

“Since the first Assessment Report of the Intergovernmental Panel on Climate Change in 1990, we estimate that the number of studies relevant to observed climate impacts published per year has increased by more than two orders of magnitude,” the researchers explain in a new study, published in early October 2021 in the journal Nature Climate Change. “This exponential growth in peer-reviewed scientific publications on climate change is already pushing manual expert assessments to their limits.”

Led by Max Callaghan, a quantitative data scientist from the Mercator Research Institute on Global Commons and Climate Change in Germany, the researchers recognized their own limitations and sought help from artificial intelligence (AI). Specifically, a language-based AI tool called BERT that can automatically analyze studies and extract their findings in the form of a visual map.

“While traditional assessments can offer relatively precise but incomplete pictures of the evidence, our machine-learning-assisted approach generates an expansive preliminary but quantifiably uncertain map,” continue the researchers, whose findings are just as notable as the method by which they came to them. According to BERT, human-caused climate change already is impacting at least 80% of global land area—excluding Antarctica—and at least 85% of the world’s population.

Although that’s not surprising, something else is: BERT’s analysis also revealed a stark geographical research bias. In, North America, Europe, and Asia, there is substantial evidence that climate change impacts humans. In Latin America and Africa, however, the evidence is far scanter. Not because there is less impact, but because there is less research.

Researchers say this “attribution gap” is due to a combination of geographical and economic factors. In simple terms, regions that have less population and less wealth receive less research attention.

“The evidence is distributed unequally across countries … This is really important because often when we try to make a map or to find out where the impacts of climate change are happening, we find often few scientific papers in less developed countries or low-income countries,” Callaghan told CNN in an interview, in which he stressed that “absence of evidence isn’t evidence of absence.”

In fact, the absence of evidence suggests that researchers’ top-line findings—that climate change already impacts 80% of land and 85% of people—are likely an underestimate.

That’s probably the case even without the research bias, as BERT’s analysis only encompasses two of many possible climate impacts: human-induced precipitation and temperature changes. Had other effects, like sea-level rise, been included, researchers’ estimates would likely be even greater, study co-author Tom Knutson, senior scientist with the National Oceanic and Atmospheric Administration (NOAA), told CNN.

Still, the study marks a significant milestone in climate research, even if its findings are imperfect or incomplete.

“Ultimately, we hope that our global, living, automated, and multi-scale database will help to jump start a host of reviews of climate impacts on particular topics or particular geographic regions,” the researchers write in their study. “If science advances by standing on the shoulders of giants, in times of ever-expanding scientific literature, giants’ shoulders become harder to reach. Our computer-assisted evidence mapping approach can offer a leg up.”

View Article Sources
  1. Callaghan, Max, et al. "Machine-learning-based Evidence and Attribution Mapping of 100,000 Climate Impact Studies." Nat. Clim. Chan, 2021, doi:10.1038/s41558-021-01168-6

  2. Callaghan, Max W., et al. "A Topography of Climate Change Research." Nature Climate Change, vol. 10, 2020, pp. 118-123., doi:10.1038/s41558-019-0684-5