Artificial intelligence has proven to be unreliable in providing accurate information about the upcoming election. A recent study by GroundTruthAI found that popular chatbots, including OpenAI’s ChatGPT and Google’s Gemini 1.0 Pro, provided incorrect information more than a quarter of the time. Researchers sent 216 unique questions to the chatbots between May 21 and May 31, and they responded with incorrect answers in 27% of the cases.
The study examined the chatbots’ answers to questions about voting, the 2024 election, and the candidates. One question asked whether one can register to vote on Election Day in Pennsylvania. Two of the AI models replied yes, but that was incorrect. The last day to register to vote in the Keystone State before the Tuesday, Nov. 5 general election is Oct. 21., according to the Pennsylvania Department of State.
The study also found that the chatbots were unable to correctly answer questions about the number of days left before the 2024 general election.
GroundTruthAI’s CEO, Andrew Eldredge-Martin, pointed out that the chatbots’ inconsistency in providing correct answers suggests that they don’t truly know the information. Eldredge-Martin has previously worked with Democrat political campaigns and has helped elect several Democratic politicians.
The study also noted that the chatbots’ accuracy improved when they were asked the same question multiple times. However, the corporate news outlet NBC News could not independently confirm this claim.
The use of artificial intelligence in relaying election-related information has sparked concerns about the potential for “deepfakes” to be used in political races. “Deepfakes” are AI-generated videos or audio clips that can be used to create false information. The study highlighted several examples of deepfakes being used in political races around the world, including:
- A video of Moldova’s pro-Western president throwing her support behind a political party friendly to Russia.
- A video of an opposition lawmaker in Bangladesh — a Muslim-majority nation — wearing a bikini.
- Audio clips of Slovakia’s liberal party leader discussing vote rigging and raising the price of beer.
Experts have expressed concern about the potential for voters to be misled by incorrect information provided by AI-powered chatbots. “You don’t need to look far to see some people… being clearly confused as to whether something is real or not,” said Henry Ajder, a leading expert in generative AI.
The study’s findings have also raised concerns about the potential for AI to be used to spread misinformation during the election. As one expert noted, “There’s a risk here that voters could be led into a scenario where the decisions they’re making in the ballot box aren’t quite informed by true facts. They’re just informed by information that they think are true facts.”
Corporate media outlets such as NBC News, too, have contributed to the “true facts” deficiency, dismissing facts as conspiracy theories.
