Donald Trump’s presidency made clear how much of our public discourse has shifted this century — away from traditional media and face-to-face exchange and onto social media platforms. Every second, 6,000 tweets are posted on Twitter, where we divulge aspects of our lives – our desires, emotions, thoughts and political opinions – in 280 characters or fewer.
But what may be underappreciated is how online communication has altered not only what we communicate about our leaders, but how we communicate to them. We are already seeing this in the first weeks of the Biden administration.
On Jan. 20, 2021, a few moments before a historic inauguration, Joseph Robinette Biden, Jr. tweeted, “It’s a new day in America.” Around noon on the same day, he was sworn in as the 46th president of the United States of America, proclaiming “Democracy has prevailed” in his first address to the nation.
Since that time, the 46th president has faced calls from ordinary citizens on what he should do as commander in chief — a development lost in the 24/7, “what’s trending now” lens that typifies our understanding of social media content.
From Jan. 20 to Jan. 29, we collected a sample of about 300,000 tweets to understand what Twitter users suggest the new president should do. We fed tweets that contain key phrases related to “President Biden should” to our natural language understanding algorithms – a form of Artificial Intelligence (AI) that enables computers to analyze large amounts of text data – before filtering and reading sample tweets to remove noise whenever possible. Our goal was to mine the wisdom of online crowds who expressed their views and wishes in tweets with phrases semantically similar to “President Biden should.”
Many thousands of tweets indicated and supported the idea that the president should declare climate change as a national emergency (61 percent of the tweets studied), consider reinstatement of Lt. Colonel Vindman (27.5 percent) and pass and sign the $1.9 trillion COVID-19 relief bill (6 percent), in that order. There were hundreds of tweets suggesting that the president start “from scratch” on vaccine distribution (3 percent), classify domestic terrorists as terror organizations (1.5 percent) and ban assault weapons (1 percent).
In the same experiment, we tracked tweets that started with phases similar to “President Biden will” to mine what the Twitterverse thinks the president will do. About 36 percent of the experiment’s tweets were about reuniting refugee families, and another 35 percent were about halting the construction of the border wall. The third most frequent theme in those tweets was about raising taxes (18.5 percent), followed by providing a path to citizenship to undocumented immigrants (10.5 percent). Since his first day in office, President Biden has signed several executive orders and his administration has acted swiftly on several issues, the majority of which are consistent with the wishes expressed in the Twitterverse.
Over the last few years, a constellation of new alternative Big Data sources, when analyzed with AI, proved to have potential in predicting presidential elections. In fact, data of merchandise supporting candidates such as online searches of yard signs combined with tweets signaled ahead of time who was leading in the 2020 presidential elections.
Ultimately, Big Data and AI have the potential to provide a competitive edge not only to win and predict presidential elections, but also to be applied on a grander scale in order to gauge public opinion about government policies, which could help lawmakers make more data-driven decisions — and to have a better understanding of the nature of our civic engagement.
The applicability of specialized Artificial Intelligence in the world of politics is becoming more promising, and in the near future, AI will set a new paradigm towards a more data-driven world of politics.
Anasse Bari is a professor of computer science at New York University, where he teaches computer and data science courses. Prof. Bari leads the Predictive Analytics and Artificial Intelligence lab, where his team researches the strategic value of AI and Big Data to address high impact social problems. He is the co-author of the book “Predictive Analytics for Dummies.” Follow him on Twitter @BariAnasse.