Tim Cook ‘Not 100 Percent’ Sure Apple Can Stop AI Hallucinations

Even Apple CEO Tim Cook isn’t sure the company will be able to stop AI hallucinations entirely. In an interview with The Washington PostCook said he would “never claim” that his new Apple Intelligence system will not generate false or misleading information with 100% confidence.

“I think we did everything we knew how to do, including thinking deeply about the readiness of the technology in the areas where we use it,” says Cook. “So I’m confident it will be of very high quality. But I would say in all honesty that this falls short of 100 percent. I would never say it’s 100 percent.”

Apple unveiled its new Apple Intelligence system during the Worldwide Developers Conference on Monday, which will bring AI capabilities to iPhone, iPad and Mac. These features will allow you to generate email replies, create custom emojis, summarize text and more.

Apple also announced that it is partnering with OpenAI to incorporate ChatGPT into Siri. The voice assistant will turn to ChatGPT when it receives a question that is more suitable for the chatbot, but will ask for your permission before doing so. In the feature demo shown during WWDC, you can see a disclaimer at the bottom of the response that says: “Please check important information for errors.”

When asked about the integration, Cook said Apple chose OpenAI because the company is a “pioneer” in privacy and currently has “the best model.” Apple may also not only partner with OpenAI in the future. Cook responded, “We’re integrating with other people as well.” During a post-keynote live session on Monday, Apple senior vice president Craig Federighi said that Apple could eventually bring Google Gemini to iOS as well.