Apple Says Its New Google-Infused AI Is All About Privacy

Source Domain: gizmodo.com

Apple just wrapped up the opening keynote of its annual Worldwide Developer Conference (WWDC), and as expected, AI was a major focus.

At the heart of Apple’s AI news was what it described as “a big step forward” for Apple Intelligence, the company’s suite of on-device AI features first unveiled in 2024. Speaking via a livestream of the keynote, Apple senior vice president of software engineering Craig Federighi said Apple Intelligence had been upgraded with a “bold new architecture,” making apps “smarter and more useful,” including a revamped Safari and a long-awaited Siri upgrade called Siri AI.

Federighi said that the new architecture is powered by foundation models codeveloped by Apple and Google, via the company’s strategic partnership, which kicked off in January. Those models can run directly from Apple devices as well as from private cloud servers, Federighi noted, and unlock “a huge upgrade for Apple intelligence, with state-of-the-art understanding and reasoning, and multiple modalities like powerful image understanding and generation.”

Also beneath the hood of the new architecture is a system orchestrator, designed to boost the AI system’s contextual awareness for each user by pulling personal information from the apps they use on Apple devices. It’s also capable of “broad world knowledge”—technobabble for being able to access the web and respond to user queries with up-to-date information.

We’re not like other AI companies

Apple also played up its commitment to privacy with its latest AI upgrades, describing it as “non-negotiable.”

During the WWDC keynote, Federighi tried to separate Apple from other AI developers, whom he said may “talk about privacy, but by default, most of them retain your personal interactions, leaving the onus on you to defend your privacy…” The fact that the new Apple Intelligence architecture runs either using device-specific or private cloud compute, he said,…

Source