Only 14% of adults are even somewhat likely to use AI to get information about the presidential election, and there is a bipartisan consensus that the use of AI by either voters or candidates would be more of a bad thing than a good thing, according to a new study from the University of Chicago Harris School of Public Policy and The Associated Press-NORC Center for Public Affairs Research.
AI, which is designed to be capable of performing tasks that humans can do such as recognizing speech or pictures, is a relatively unfamiliar technology for most people. Overall, 54% of adults have not read or heard much at all about AI and just 30% report they have used an AI chatbot or image generator. Although younger adults are more familiar with AI, they are no more likely than older adults to say they will use AI chatbots to get information about the election.
The public largely agrees that the use of AI by either voters or candidates would be more of a bad thing than a good thing, although they are split on using AI chatbots to find information about how to register to vote. Regardless of partisanship, fewer than 1 in 10 say it would be a good thing for voters to use chatbots to decide who to vote for, for candidates to tailor political advertisements to individual voters, or for candidates to use AI to edit or touch-up photos and videos.
Adults are more likely to say it is a good thing for voters to use AI chatbots to find information about how to register to vote (37%) or information on how to cast a ballot (26%).
“The survey shows people believe AI can be best used to find information about how to register to vote or cast a ballot, which is the type of technical information that AI researchers are most concerned about chatbots providing,” said Ethan Bueno de Mesquita, the Interim Dean and Sydney Stein Professor at the University of Chicago Harris School of Public Policy. “If voters are asking chatbots how to register to vote or where they can cast a ballot, there are concerns that chatbots might just make up an answer. Fortunately, this is also an area where AI companies seem attuned to making sure chatbots direct users toward more authoritative information.”
False information
The public is worried about the role AI could play in spreading false election information. Most adults (58%) believe that the use of AI will increase the spread of misinformation during the 2024 presidential election, and those who are more familiar with AI tools are more likely to say its use will increase the spread of misinformation (66% vs 51%).
The survey reveals that about 8 in 10 adults believe that technology companies, social media companies, the news media, and the federal government all share at least some responsibility when it comes to preventing the spread of misinformation by AI.
There is bipartisan support when it comes to a variety of actions to address the use of AI in the election. Majorities favor the federal government banning AI-generated content that contains false or misleading images in political ads (66%), technology companies labeling all AI-generated content made on their platforms (65%), and politicians and the groups that support them pledging not to use AI-generated content in their campaigns (62%).
“The findings highlight that the public is very skeptical about any positive impacts from AI being used in the 2024 presidential elections,” said David Sterrett, a senior research scientist at NORC. “Given all the concerns about AI, majorities of both Democrats and Republicans seem open to various actions to regulate the use of AI ahead of the 2024 election.”
About the study
This study was conducted by the University of Chicago Harris School of Public Policy and The Associated Press-NORC Center for Public Affairs Research. Staff from Harris Public Policy and The AP-NORC Center collaborated on all aspects of the study. Interviews for this survey were conducted between October 19 to 23, 2023, with adults age 18 and older representing the 50 states and the District of Columbia. Panel members were randomly drawn from AmeriSpeak, and 1,017 completed the survey. Interviews were conducted in English. The overall margin of sampling error is +/- 4.1 percentage points at the 95 percent confidence level, including the design effect.
This story was adapted from the Harris School of Public Policy website.
—Prof. Jacqueline Stewart
