As if worrying about fighting fake news was not enough, there is more and more information (from scientists) about better ways to deal with coronavirus. The phenomenon, dubbed "infodemic" by the World Health Organization, has made it very difficult for researchers (and not only) to fully assimilate rapidly evolving discoveries, making some current research obsolete even before being evaluated by other scientists.
Research in recent months has been very demanding and researchers consider it their duty to publish results that may be useful to clinicians. But there are always conflicts due to the growing scientific literature.
In a opinion article In Patterns magazine, Ganesh Mani of Carnegie Mellon University, an investor, technology entrepreneur and assistant professor at the Software Research Institute, and Tom Hope, a postdoctoral researcher at the Allen Institute for AI, issued a distance call.
"Given the ever-increasing volume of research, it will be difficult for people to keep up," the article said.
They report in particular the flood of research information on coronaviruses. As of mid-August, more than 8.000 drafts of Covid 19-related scientific papers had been published in online medical, biological and chemical records. There is not a lot of information about depression caused by quarantine. In the field of virology, the average time used for peer review and the publication of new articles decreased by an average of 117 days, which was initially 60 days.
So it seems more and more attractive and perhaps necessary to combine human know-how with AI to start helping record results with research that leaps and bounds. Too much information not only leads to the impossible digestion of all, but also to the distinction between useful and suspicious information and results. Artificial intelligence could help evaluate research and classify it appropriately.
"We will have the same discussion with vaccines," said Mani. "We will have a lot of discussions."
Of course, technology alone cannot find a complete solution. Mani and Hope are proposing new policies, such as highlighting the negative effects of positive findings, which may be important for clinicians as they discourage scientists from accessing limited or unnecessary research. Other ideas presented in the article include identifying top quality reviews and linking research to relevant literature, recall sites, or legal decisions.
Artificial intelligence could help, but there is still a problem in understanding human language. Thus the authors state that it may be necessary for researchers to write two versions of research papers, one for humans and one for AI.
"Using such infrastructure will help society in the next big surprise or challenge, which is likely to need as much, if not more, knowledge."