Apple may never be able to get its AI to stop lying, CEO says

Apple Chief Executive Tim Cook says the company’s latest artificial intelligence technology, known as “Apple Intelligence,” could be a big liar.

This admission, made by Mr. Cook in a recent interview with the Washington Post, highlights a common problem plaguing AI chatbots in the tech industry.

During the interview, Cook was asked about his confidence in Apple Intelligence’s ability to avoid generating false information, a phenomenon often referred to as “hallucination” in the AI ​​world. Mr Cook acknowledged that there are still many uncertainties.

“It’s not 100 percent,” he said. While expressing confidence in the overall quality of the technology, he added: “I would honestly say that we are far from 100%. I would never claim it’s 100 percent.

The revelation is particularly concerning given the intense attention and investment in AI technology from the tech industry and Wall Street. Despite the billions of dollars invested in AI development, the tools often produce false or misleading information.

In some cases, AI models were trained on unreliable data, leading to even more inaccuracies. For example, Google’s AI-powered search recently advised a user to put glue on their pizza, referencing an old Reddit joke.

Mr. Cook is not the only one with his concerns. Google Chief Executive Sundar Pichai recently made similar comments, acknowledging that while progress has been made, the problem of AI-generated lies is far from resolved.

“We’ve definitely made progress when we look at reality metrics year over year,” Mr. Pichai said in an interview with The Verge. “We’re all making things better, but it’s not fixed.”