Artificial Intelligence Can Sense the Political Divide of COVID-19, Study Finds


OSU researchers created AI that can distinguish US Congresspeople along political party lines based on Tweets.

The coronavirus pandemic produced political chaos, yielding confusion about how to behave appropriately in the current climate. 

AI can distinguish US Congresspeople along political party lines based solely on the text and date of their Twitter messages. Researchers at Ohio State University developed an algorithm that could correctly identify a member’s political affiliation 76% of the time.

The study, published on Wednesday in the journal entitled “Science Advances,” found that Democrats’ tweets frequently discussed “threats to public health and American workers, while Republicans placed greater emphasis on China and businesses.”

“It is remarkable that we could identify partisanship even when members have only 280 characters to send their messages on Twitter,” said Skyler Cranmer, Carter Phillips and Sue Henry Professor of Political Science at Ohio State University and a co-author of the study.

The researchers explained: “The severity of this crisis is particularly sensitive to public opinion given that behavioral change at the individual level is integral to successfully slowing the spread of the virus. Given the high levels of polarization in the American electorate, citizens are less likely to change their behavior in ways that correspond to the consensus of public health experts if there is not a political consensus that such changes are necessary.”

Researchers pointed out the political consensus in response to the terrorist attacks on September 11, 2001, “when Republican and Democratic lawmakers issued joint statements reassuring Americans they were safe and promising rapid retaliation.”

The political divide of COVID-19 evolved over the months. The algorithm analyzed all 30,887 tweets that US congresspeople wrote about the virus from January 17 to March 31. Artificial intelligence was slow to accurately distinguish Democrats from Republicans as authors of tweets from the first week, which “indicates there was little polarization.” When the first infectious COVID-19 case was confirmed in California, the algorithm improved at separating the politicians’ tweets along party lines. 

When President Trump declared the pandemic a national emergency on March 13, “parties debated the various relief packages designed to mitigate the economic damage caused by” COVID-19. Following this declaration, the party divide widened: Democrats posted “significantly” more COVID-related tweets than Republicans (19,803 vs. 11,084). 

“This suggests Democratic members were sending earlier and stronger signals to their constituents that they should be concerned about the crisis,” Cranmer said.

The study found that Democrats’ tweets often included the words “health” and “leave,” regarding aid for workers, and “testing.” Republicans frequently tweeted the words “together,” “United States,” “China,” and “businesses,” emphasizing “a general need for national unity” and “[framing] the pandemic as a war.”

The AI categorized only around 31% of the members in the “partisan overlap” bucket when the algorithm could not distinguish party alliance based on the tweets alone.

“That means for 69% of members, their tweets are more partisan than the most similar member of the other party,” said Jon Green, co-author and doctoral student in Political Science at Ohio State.

“In democratic countries, the public is highly responsive to cues sent by political elites whose messages can encourage unity or deepen social cleavages. Because the public relies on these cues for reliable information, it is especially important that elites present a unified message during a crisis.”

Read the full report.


Science, Futurology, and Analysis