Some problems are so big, you can’t really see them.
Climate change is the perfect example. The basics are simple: the climate is heating up due to fossil fuel use. But the nitty gritty is so vast and complicated that our understanding of it is always evolving. Evolving so rapidly, in fact, that it’s basically impossible for humans to keep up.
“Since the first assessment report (AR) of the Intergovernmental Panel on Climate Change (IPCC) 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,” scientists explain in a new paper, led by first author and quantitative data researcher Max Callaghan from the Mercator Research Institute on Global Commons and Climate Change (MCC) in Germany.
“This exponential growth in peer-reviewed scientific publications on climate change is already pushing manual expert assessments to their limits.”
This struggle is its own problem, of course, because how can humans ever grasp the problem of climate change, if the size of the problem defies our ability to objectively analyze it, measure it, and understand it?
One solution to this ‘big literature’ dilemma calls for a very different kind of entity doing the reading – using artificial intelligence (AI), rather than humans, to sift through the almost limitless and ever-expanding mountain of published climate science.
In their new study – yes, another one to add to the list – Callaghan and co. did just that, using a deep-learning language analysis AI tool called BERT to identify and classify over 100,000 scientific studies detailing the impacts of climate change.
While the researchers acknowledge that automated analyses like this are no substitute for the careful assessments of human experts, at the same time, their method can do things human’s simply can’t.
In this case, that meant crunching vast amounts of data, identifying a huge range of different kinds of climate impacts, mapping them out across every continent, and interpreting them in the context of anthropogenic contributions to historical temperature and precipitation trends.
We need to be careful with it, though, because machine-learning analyses like this – especially at such staggering scale – can contain false positives and other kinds of uncertainties, the researchers say.
“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,” the researchers write.
Before that, however, the AI analysis has already generated some troubling statistics.
According to the study, 80 percent of global land area (excluding Antarctica), already shows trends in temperature and/or precipitation that can be attributed at least in part to human influence on the climate – and these climate impacts already touch an estimated 85 percent of the world’s population.
Of course, we didn’t need any artificial superbrain to tell us that climate change was a giant problem, but what’s telling is where climate impacts can and can’t be clearly discerned – based on where studies have been geographically focused.
For around half (48 percent) of the world’s land – hosting three quarters (74 percent) of the global population – high levels of evidence of impacts on human and natural systems were co-located with attributable temperature or precipitation trends.
In other words, in places like western Europe, North America, and South and East Asia, there’s a lot of overlap between impacts on the natural world and research into human-caused contributions to climate change.
In other places, however, the links aren’t as strong – but maybe only because, ironically enough, there’s not enough climate science yet looking into those specific regions.
“The lack of evidence in individual studies is because these locations are less intensively studied, rather than because there is an absence of impacts in these areas,” the researchers suggest, noting this “attribution gap” is due to both geographic characteristics (inhospitable or sparsely populated areas) and economic considerations (low-income countries are significantly less studied).
“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 team concludes.
“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.”
The findings are reported in Nature Climate Change.