Study: Chatbots' inaccurate election info may deter voters

A new study conducted by AI Democracy Projects and Proof News has revealed that AI-powered chatbots, including popular ones like GPT-4 and Google's Gemini, are producing inaccurate and harmful election information more than half the time. As the U.S. presidential primaries are underway, experts are warning that these chatbots could potentially discourage voters from going to the polls or mislead them with false information.

The study tested five AI models, including OpenAI's ChatGPT-4 and Meta's Llama 2, and found that they often provided incorrect responses to prompts. For example, Llama 2 erroneously stated that California voters can vote by text message, which is not legal in the U.S. Additionally, none of the AI models correctly stated that wearing campaign logos at Texas polls is prohibited under state law.

Policy experts believe that AI could be beneficial in improving elections, such as by powering tabulators that can scan ballots quickly. However, there are concerns that bad actors, including governments, could misuse AI to manipulate voters and weaken democratic processes. For instance, AI-generated robocalls were used to discourage voters from participating in the New Hampshire presidential primary.

Furthermore, Google recently paused its Gemini AI picture generator after it produced historically inaccurate information. The chatbot provided racially diverse images when asked to create an image of a German soldier during World War 2, which raised questions about the accuracy of AI models.

Despite the widespread fear among adults in the U.S. that AI tools will spread false information during this year's elections, Congress has yet to pass laws regulating AI in politics. This lack of regulation leaves tech companies responsible for governing themselves when it comes to the use of AI in elections. With Super Tuesday approaching and millions of Americans relying on chatbots for election information, the accuracy and reliability of these AI tools are under scrutiny.