Capstone Project Seeks to Improve Voter Turnout with AI
Featuring Award-Winning MIDS Summer 2018 Capstone Project “ElectBot AI”
By many standards, voter turnout in the United States is low, and not just for young people. The U.S. trails most developed countries in voter turnout, as demonstrated by recent Pew Research Center studies. A group of Masters of Information and Data Science (MIDS) students at the School of Information recognized this problem and set out to build a solution.
Electbot AI is a chatbot built with artificial intelligence to tackle this problem of low voter turnout in US elections, by performing the essential functions of informing voters and reminding them to vote.
ElectBot AI began as a MIDS Capstone Project for Summer 2018 created by recent alumni Ben Attix, @Krista Mar, and Prajakta Pandharkar. The chatbot chats with potential voters through SMS to ensure ElectBot AI is accessible to all people and to break through the noise on other channels; informs potential voters about candidates’ stances on issues that are important to them and provide crucial information on how to vote; and, reminds potential voters to vote the day before an election. As stated on their website, Electbot AI “provides crucial information about voting, which makes it easier for first-time or infrequent voters to be informed.”
This summer, the ElectBot AI team partnered with vote.org for the Michigan midterm election primaries on August 7. Vote.org provided opt-in phone numbers ~56,000 potential voters in Michigan, and they were able to contact over 6,000 of those with ElectBot. “Partnering with Vote.org allowed us to reach many more users than we would have been able to otherwise,” said the Krista Mar. The team is still analyzing the results of the experiment as they receive all the voter turnout data. They plan to publish the results on vote.org.
“We saw ElectBot as a way to democratize information and provide easy to access information to vote.”
“We saw ElectBot as a way to democratize information and provide easy to access information to vote,” said Mar. “Our working hypothesis was that people will vote if they understand candidates’ stances on issues that affect their lives (e.g. abortion rights, immigration, health care), and have easy access to information about how to vote or where to vote. By building a chatbot to deliver these stances in an interactive matter, we hoped to decrease the barrier to finding information about candidates and increase people’s awareness about the candidates.”
According to the capstone team, the initial inspiration for ElectBot came from Mar’s first experience voting. The first time she voted at a polling station, she knew the major candidates in the presidential election but didn’t know about the smaller, state, or local candidates, so she ended up just voting along party lines. Discouraged by this experience, Mar attempted to do better research before voting in subsequent elections but found it difficult to navigate the information and adequately compare candidates and issues.
Additionally, as the team was working on Electbot, Mar was also working philanthropically on the issue of civic engagement in communities of color. As a MIDS student, she was naturally thinking about how data science could be utilized to improve voter turnout in marginalized communities.
Moving forward, vote.org has demonstrated an interest in partnering with ElectBot in future elections or with other civic engagement efforts. Through conversations about ElectBot, other organizations have also approached the team to discuss working on other projects including a “know your rights” campaign for immigrant workers. “We are still in the beginning stages of this conversation,” said Mar, “but hope to continue to use our data science skills to solve problems with social impact!”
Demo ElectBot AI yourself, and learn more on Electbot.org.
UPDATE, January 12, 2019: ElectBot AI was awarded the Hal R. Varian Capstone Award for outstanding MIDS projects for Summer 2018!