By Emmanuel Ogbodo
Meta encountered an issue when its AI assistant wrongly claimed that the recent attempted assassination of former President Donald Trump didn’t happen. A company executive has attributed this mistake to the AI technology powering its chatbot.
In a blog post on Tuesday, Joel Kaplan, Meta’s global head of policy, called the AI’s responses to questions about the shooting “unfortunate.” Initially, Meta programmed the AI not to respond to queries about the attempted assassination, but the restriction was lifted after people noticed. Kaplan acknowledged that “in a small number of cases, Meta AI continued to provide incorrect answers, including sometimes asserting that the event didn’t happen – which we are quickly working to address.”
“These responses are known as hallucinations, an issue affecting all generative AI systems, posing ongoing challenges in handling real-time events,” Kaplan said. “Like all generative AI systems, models can produce inaccurate or inappropriate outputs. We will continue to address these issues and improve these features as they evolve and more people provide feedback.”
Meta isn’t alone in this challenge. On Tuesday, Google also had to deny allegations that its Search autocomplete feature was censoring information about the assassination attempt. Trump reacted on Truth Social, saying, “Here we go again, another attempt at RIGGING THE ELECTION!!! GO AFTER META AND GOOGLE.”
Since ChatGPT’s debut, the tech industry has been dealing with the issue of generative AI’s inaccuracies. Companies like Meta have tried to use high-quality data and real-time search results to counteract hallucinations. However, as this case illustrates, overcoming the tendency of large language models to fabricate information remains difficult.
