Perplexity rolls out an AI election information hub. Use it at your own risk

While most major AI services, including OpenAI and Google’s Gemini, are staying quiet about the election, Perplexity is going all in on the 2024 presidential race. The AI service has set up a hub that offers background and, starting Tuesday, purported real-time information about the results. But you might want to think twice before using it.

The Election Information Hub, as it’s called, is designed to help voters understand key issues and vote intelligently (and later track results). The idea is a solid one: Cut through the clutter to let people know where candidates stand and some pros and cons of ballot issues. However, as with many artificial intelligence ventures, it’s still far from perfect, which could result in voter confusion instead.

A quick check of the tool found ballot measures in some districts not being mentioned and no notification that Robert F. Kennedy Jr. had dropped out of the presidential race (though the site has reportedly updated his information page to reflect that in some, but not all, summaries).

Some of the information sources that Perplexity is using are unusual—and sometimes questionable as well. The Election Information Hub regularly cites Wikipedia, which can be edited by anyone at any time. (Even Wikipedia says it is “not a reliable source for citations” on its own platform.) Other cited sources include IMDb, Libcom.org (a far-left leaning outlet), Major League Baseball, and X.com.

(Perplexity did note information on the “Your Ballot” tool is powered by the Elections API from the nonprofit Democracy Works—and pulled directly from that group. Real-time results on Tuesday will use data from the Associated Press, the company said.)

The hub also uses information from Forbes (which, in June, threatened Perplexity with legal action over copyright infringement) and the New York Times (which sent a “cease and desist” notice to the company three weeks ago over content use). Perplexity CEO Aravind Srinivas also posted about his company’s election hub on X, as the New York Times Tech Guild strike appeared to threaten access to the publication’s interactive election features.

Hey AG Sulzberger @nytimes – sorry to see this. Perplexity is on standby to help ensure your essential coverage is available to all through the election. DM me anytime here. https://t.co/SC1vCZlBtw

In a blog post, Perplexity (which was founded with backing from Amazon founder Jeff Bezos two years ago) positioned the hub as a tool to receive “trusted, easy to understand information to inform your voting decisions,” but did not caution users about potential gaps or bad information.

Asked about how it was working to avoid AI hallucinations, a company spokesperson told Fast Company, “Perplexity uses a process called Retrieval-Augmented Generation to identify relevant information and summarize it in a way that’s tailored to a user’s query. For election-related questions, we are using a curated set of sources that are nonpartisan and fact-checked (such as Ballotpedia and news organizations). We’re actively monitoring our systems to ensure that we continue to prioritize these sources when answering election-related queries. Unlike other AI chatbots, answers on Perplexity are not utilizing stored knowledge from a model’s training data, which allows us to minimize hallucinations.”

The launch of the Election Information Hub comes just days after a slew of warnings from government officials about using chatbots as an information source in this year’s election. New York State Attorney General Letitia James, in a statement, said voters who relied on a chatbot risk being misinformed and could even potentially miss out on their opportunity to vote if they follow inaccurate information about voting times and locations.

“My office’s testing of AI chatbots found many providing false or misleading information about voting, threatening New Yorkers’ ability to exercise their right to vote,” James said. “With early voting underway and Election Day just around the corner, I urge voters to be cautious and seek reliable information about the election only from official sources.”

Alexandra Reeve Givens, CEO of the Center for Democracy & Technology, backed up that thinking, telling CNBC, “Voters categorically should not look to AI chatbots for information about voting or the election—there are far too many concerns about accuracy and completeness.”

And a Pew Research poll in September found more than half of Americans worried about the impact of AI on the 2024 election.

That may be why many other AI chatbots are refusing to be drawn into election conversations. ChatGPT tells users who ask about the 2024 race: “I can’t provide updates on specific political situations or candidates. I recommend checking reputable news sources for the latest information on the presidential election.” Gemini defers to Google’s chief business, saying “I’m trained to be as accurate as possible but I can make mistakes sometimes. While I work on perfecting how I can discuss elections and politics, you can try Google Search.”

AI has a spotty history already in this election cycle. Grok, X’s AI chatbot, at one point, incorrectly told users it was too late for Kamala Harris to take Joe Biden’s spot on the ballot after Biden dropped out of the race. And in February, an AI Democracy Projects investigation tested five chatbots (though not Perplexity), finding half of the responses to election queries were inaccurate and 40% were deemed harmful.