Apple May Never Be Able To Make Its AI Stop Lying, CEO Says
Despite billions of dollars poured into AI development, the tools often produce false information.
The chief executive of Apple, Tim Cook, says that the companyâs latest artificial intelligence technology, known as âApple Intelligence,â may be a big fat liar.
The admission, which Mr. Cook made in a recent interview with the Washington Post, highlights a common issue that has plagued AI chatbots across the tech industry.
During the interview, Mr. Cook was asked about his confidence in Apple Intelligenceâs ability to avoid generating false information, a phenomenon often referred to as âhallucinationâ in the AI world. Mr. Cook acknowledged that there are still many uncertainties.
âItâs not 100 percent,â he said. While he expressed confidence in the overall quality of the technology, he added, âIâd say in all honesty thatâs short of 100 percent. I would never claim that itâs 100 percent.â
The revelation is particularly concerning given the intense focus and investment in AI technology by both the tech industry and Wall Street. Despite billions of dollars being poured into AI development, the tools often produce false or misleading information.
In some cases, AI models have been trained on unreliable data, leading to further inaccuracies. For example, Googleâs AI-powered search recently advised a user to put glue on their pizza, referencing an old joke from Reddit.
Mr. Cook is not alone in his concerns. Googleâs chief executive, Sundar Pichai, recently made similar comments, acknowledging that while progress has been made, the issue of AI-generated falsehoods is far from resolved.
âWe have definitely made progress when we look at metrics on factuality year on year,â Mr. Pichai said in an interview with The Verge. âWe are all making it better, but itâs not solved.â