Apple's AI chief gave this warning to the Siri team when he started working with them

Siri struggled when it launched 14 years ago, but when John Giannandrea joined as Apple's AI chief in 2018, he said that failure wasn't an option.

• Apple's head of AI strategy and machine learning, John Giannandrea, has run Siri since 2018.
• His first instruction to the Siri team, he says, was "failure is not an option."
• The popularity of Apple's voice assistant makes it an exceptional proving ground for AI.

When Apple's head of AI strategy and machine learning, John Giannandrea, joined the company in 2018, the company's voice assistant, Siri, was struggling to live up to its promise.

Looking for a turnaround, Giannandrea says his first instruction to the Siri team was inspired by legendary NASA Chief Flight Director Gene Kranz: "Failure is not an option."

"A lot of people use Siri a lot of the time," Giannandrea said in an interview with John Gruber after Apple's WWDC last week. "And as it's gotten better over the years, we see in our data that people just use it more."

Now 14 years after Apple launched Siri, the voice assistant's growing popularity has made it an exceptional proving ground for AI, raising the stakes still further.

The recent introduction of Apple Intelligence, as well as a partnership with OpenAI, may bring a flood of new users to Siri, each with more complicated requests than ever before.

And given the central role of Apple's iPhone, it's fitting that Giannandrea is leading this effort. He has long argued that on-device AI models are critical to the technology being practical for everyday users.

"The inference of large language models is incredibly computationally expensive," he said, explaining why Apple's newest A17 chips are so well suited for the task. "It's the oomph in the device to actually do these models fast enough to be useful. You could in theory run these models on a very old device, but it will be so slow will not be useful."

In addition to getting the technical parameters right, Giannandrea said his team is also wrestling with philosophical and ethical questions about the role AI will play in people's lives.

"We've been very careful about applying this technology in a very thoughtful way," he said. "We've tried to corral this technology to do what it's really good at doing."

At the same time, even though researchers are concerned about "the safety problem," Giannandrea said Apple is not interested in limiting the creativity of its users with an overly restrictive set of rules.

"It's a very fine balance as we spend a lot of time and a lot of years trying to figure out because this is new for us. This is new for the whole industry," he said.