Twitter ‘Bots’ Becoming More ‘Human,’ Will Influence 2020 Presidential Election, New Research Study Says

New research at the University of Southern California shows that automated Twitter accounts, known as 'bots,' are becoming more effective at simulating human behavior online.

Donald Trump holds a campaign rally in Rochester, New York.
Brett Carlsen / Getty Images

New research at the University of Southern California shows that automated Twitter accounts, known as 'bots,' are becoming more effective at simulating human behavior online.

A new study conducted by researchers at the University of Southern California and published last week in the journal First Monday has identified approximately 31,000 Twitter accounts which posted frequent political tweets during both the 2016 presidential election and the 2018 midterm elections. However, those accounts, the researchers found, were not operated by human beings, but instead were “bots” — automated accounts that post on Twitter with no actual person behind them — that tried to influence online political discussions.

Not only did those bots overwhelmingly tweet in support of Donald Trump, but between the 2016 and 2018 elections, they grew more sophisticated, mimicking human behavior in an increasingly realistic fashion, according to a summary of the study by AdWeek. As the bots become more difficult to distinguish from actual human beings, the USC researchers say that they pose a growing threat to the 2020 presidential election — when they will likely continue to support Trump.

Trump in August claimed that online activity, particularly on Google, was biased against him, and that it somehow changed millions of votes that would have gone in his favor to votes for his 2016 opponent Hillary Clinton, as The Inquisitr reported. But most evidence, including the new research from USC, suggests otherwise, instead claiming that Trump was the beneficiary of online public opinion manipulation.

Twitter user looks at his phone.
  Leon Neal / Getty Images

“With the upcoming 2020 U.S. elections, the integrity of social media discourse is of paramount importance to allow a democratic process free of external influences,” researcher Emilio Ferrara, a USC computer science professor who served as lead author on the study.

During the 2016 election, the Trump-supporting Twitter bots engaged primarily in retweeting posts originating with other, presumably legitimate accounts, according to a Defense One report on the USC research.

Loading...

But by 2018, the researchers wrote, “bots better aligned with humans’ activity trends, suggesting the hypothesis that some bots have grown more sophisticated.” The bots greatly reduced their retweeting activity, focusing instead on posting interactive tweets, such as Twitter polls. These surveys were allegedly designed to influence political opinions online and to gather data on users who “vote” in the polls, or answer questions posed by the bot accounts.

In 2016, the bots frequently reposted the same messages over and over again — but each time from the same account. But in 2018, the distribution of identical messages was shared among multiple bots, the study found.

While the study did not draw conclusions as to how greatly the Twitter bots influenced the outcomes of the 2016 and 2018 elections, the authors warned that the growing “mutability” of bots “should be [a] cause of concern when considering the integrity of the online information ecosystem, especially with respect to online political discussions.”