Here's Why You Shouldn't Use AI to Vote
wired ·
As Americans head to the polls for the midterms, a significant portion of voters are relying on AI chatbots for candidate and policy information. However, research indicates that these tools frequently provide outdated, hallucinated, or factually incorrect data. Studies show that nearly 40 percent of election-related answers from mainstream chatbots contain at least one error, with specific examples including wrong ballot deadlines and confused biographical details. Furthermore, these AI models often cite unreliable "pink slime" websites funded by political groups rather than credible news sources. OpenAI’s ChatGPT was identified as the worst offender in citing such biased sites, while xAI’s Grok performed relatively better but still contributed misinformation. Experts warn that consumers are ill-equipped to navigate this disinformation landscape, noting that companies have overpromised on AI's capabilities. The reliance on predictive text generation means that when the AI lacks knowledge, it tends to invent answers rather than admit uncertainty.
wired