Apple Unveils Revamped 'Siri AI' at WWDC, Powered by Google's Gemini
- Timeline: From delays to WWDC reveal
- What’s new: A Gemini‑powered, app‑like chatbot
- Privacy architecture and technical trade‑offs
- Reception: Impressive, late, and cautiously optimistic
Apple Unveils Revamped ‘Siri AI’ at WWDC, Powered by Google’s Gemini Apple has rebuilt Siri as a full-fledged AI assistant, betting that tight integration and strict privacy controls can offset its late arrival in the chatbot race.
Timeline: From delays to WWDC reveal
Apple first laid out an ambitious “Apple Intelligence” vision with a smarter Siri two years ago, but many of those features never shipped, drawing criticism that the company was “behind” in AI. After a rare yearlong delay and even a $250 million class‑action settlement over undelivered Siri features, Apple returned to WWDC 2026 with what it calls an “entirely new version of Siri,” branded Siri AI.
At Monday’s keynote, Apple said Siri is now “unlocked by Apple Intelligence” and will roll out in beta later this year, initially in English and not in the EU or China. Siri AI appears across iPhone, iPad, Mac, Apple Watch, and Vision Pro, with a new design centered on the Dynamic Island and an expressive, customizable voice.
What’s new: A Gemini‑powered, app‑like chatbot
Under the hood, Siri AI is rebuilt on a custom version of Google’s Gemini model, reportedly around 1.2 trillion parameters and licensed for roughly $1 billion a year. Apple pitches it as an AI companion that can “do a lot more,” pulling from on‑device emails, messages, photos, and what’s on screen to handle multi‑step tasks like drafting messages or building packing lists from an email.
For the first time, Siri gets its own dedicated app, acting as a “warehouse” of past conversations with support for text, voice, documents, and images, synced privately via iCloud. Broader Apple Intelligence features span Safari tab management, one‑tap password updates, smarter Messages replies, and AI‑built Shortcuts workflows from plain language.
Privacy architecture and technical trade‑offs
Because Apple’s devices can’t run the largest models alone, the company introduced a three‑tier system: simple queries stay on‑device, mid‑tier tasks use Apple’s Private Cloud Compute, and the most complex requests run on Nvidia hardware in Google data centers. Apple says queries are processed statelessly and Google is contractually barred from training on Apple user data. “We believe privacy in AI is non‑negotiable,” Craig Federighi told developers.
Reception: Impressive, late, and cautiously optimistic
Hands‑on testing suggests the new Siri finally does basic but long‑requested tasks—like extracting multiple events from an email into a calendar—reliably: “AI Siri is for real this time.” Yet reviewers note the experience often feels like “Gemini, circa 2025,” with many features already standard on Android and rival chatbots.
Analysts argue Apple’s “slow‑and‑steady” approach, focused on privacy, security, and everyday usefulness, may ultimately resonate with consumers even if Siri AI is “both cool and 2 years too late.” Whether that strategy is enough to close the perceived AI gap with OpenAI, Google, and Anthropic remains an open question.
Continue reading https://foxvector.com/stories/019eaff4-ab10-39f4-713e-38632294b436
Write a comment