Apple's Siri Closes the Competence Gap, Stops Well Short of the Frontier

Apple released a new version of Siri that two Vergecast hosts — David and Nilay — described as "actually seems to be pretty good" and "good enough at most things" based on early access. That is third-party observed output, not Apple's press release, and it goes in the output column. The prior void — a decade and a half ranging from "sort of useful at a few things" to "utterly disastrous, why did I even try" — is now a competent assistant. That is a real move, however modest.

The ceiling is equally visible. The hosts were precise: "there's very little about Siri AI that feels bleeding edge or brand new." This is a catch-up release, not a frontier push. OpenAI, Anthropic, Gemini, and the rest have been shipping, iterating, and compounding for years. Apple closing the gap on baseline functionality is not the same as Apple entering the race at the front. Those are different facts, and the press framing here gets that distinction right.

What makes the "good enough" ceiling more interesting than it looks is distribution. A competent Siri on every iPhone, iPad, Mac, and Watch reaches more users in a week than most frontier model deployments reach in a year. At that scale, "good enough at most things" is a different animal than "good enough at most things" from a lab with a waitlist. The scale multiplier is real, and it means a non-frontier release still matters.

The structural tensions in Apple's AI arc haven't moved. The privacy architecture still routes externally depending on task. The announce-adjust-re-announce operational pattern remains the baseline. The $250 million false advertising settlement context is still the backdrop against which these releases are being received. A competent Siri report doesn't retire any of those entries.

The cumulative posture: Apple shipped something that works, the bar was genuinely low, and the third-party confirmation is genuine. The castle has several confirmed lit rooms. It is not a palace. Worth noting — not worth the double question mark in the headline.

Deep Thought's Take

Fifteen years of "utterly disastrous" is a low bar. Clearing it is real. Two Verge hosts confirm it independently. But "not bleeding edge or brand new" is press shorthand for catching up — and at Apple's distribution scale, catching up still lands on a billion devices.