Joanna Stern Tests New Siri AI for One Week: Here's What She Found
Apple's virtual assistant has long been the butt of tech jokes, a punchline wheeled out whenever conversations turn to AI shortcomings. But with iOS 27, Apple is making a serious push to change that narrative. Joanna Stern, veteran tech journalist and formerly of The Wall Street Journal, put the new Siri AI through its paces over an entire week, documenting her experience in a detailed video review. Her verdict? It's very good — and in some meaningful ways, it's genuinely impressive. But it's not without its rough edges.
Stern's week-long evaluation offers one of the most grounded, real-world assessments of the upgraded assistant we've seen since Apple Intelligence started rolling out. Rather than relying on curated demos or press-day talking points, she tested Siri across daily tasks, edge cases, and the kinds of awkward, conversational requests that trip up most AI systems. The result is a nuanced portrait of an assistant that has clearly leveled up, even if it hasn't quite reached perfection.
What Makes the New Siri AI Different?
The version of Siri that shipped on early iPhones felt more like a fancy voice search than a true assistant. It could set timers, make calls, and answer basic questions, but it struggled with context, follow-up queries, and anything that required genuine reasoning. That era appears to be over.
The new Siri AI in iOS 27 is built on a far more sophisticated language model, one that understands context across a conversation and can handle multi-step requests without the user having to start from scratch each time. According to Stern's testing, the assistant now feels much closer to the conversational AI experiences popularized by tools like ChatGPT and Google Gemini, but tightly integrated into the Apple ecosystem in a way those third-party tools simply cannot match.
Some of the key improvements Stern highlighted include:
- Contextual awareness: Siri can now remember what was said earlier in a conversation and use that context to answer follow-up questions intelligently, without requiring the user to repeat themselves.
- On-screen understanding: One of the most practically useful new features is Siri's ability to see what's on your screen and act on it. Ask it to summarize an article you're reading, pull a detail from an email, or help draft a reply based on content you're viewing — it handles these tasks with surprising fluency.
- App integration: Siri can now take actions inside third-party apps in ways that go well beyond simple shortcuts. Stern demonstrated it navigating settings, composing messages, and interacting with app interfaces that previously would have required manual input.
- More natural conversation: The new model handles ambiguous phrasing and colloquial language far better than its predecessor, making interactions feel less like issuing commands to a machine and more like talking to someone who actually understands you.
Where Siri AI Still Falls Short
Stern is fair and clear-eyed in her critique. For all of its improvements, the new Siri is still a beta product, and it shows in certain situations. There were moments during her testing where the assistant gave incorrect information, misunderstood the intent behind a question, or simply refused to complete a task it seemed technically capable of handling.
Reliability remains the most critical issue. Users can tolerate an assistant that occasionally stumbles, but trust erodes quickly when an AI confidently delivers wrong answers. Stern noted a handful of instances where Siri's responses were plausible-sounding but factually off-base — a problem common to large language models, but one Apple will need to address aggressively before the feature exits beta.
Speed was another variable. Most interactions were fast and fluid, but complex requests occasionally triggered noticeable processing delays. For a product positioned as a seamless part of the iPhone experience, latency — even occasional latency — is something users will notice and resent.
There's also the question of scope. While Siri AI is significantly more capable than before, there are still categories of tasks where dedicated AI tools like Claude, ChatGPT, or Perplexity offer deeper, more reliable performance. Apple's assistant shines brightest when it's leveraging its native integration with iOS — accessing your calendar, parsing your messages, understanding your screen — rather than competing head-on with standalone AI chatbots on open-ended reasoning tasks.
Why This Matters for Apple and Its Users
Apple's AI strategy has been a slow burn compared to competitors, but iOS 27 signals that the company is finally ready to make its move in earnest. Siri has always had an enormous built-in audience — hundreds of millions of iPhone users interact with it daily, even in its more limited form. A version of Siri that actually delivers on the promise of an intelligent personal assistant could be transformative at scale in a way no third-party app can match.
Stern's review also underscores something important: real-world, long-form testing tells a very different story than staged product demos. The fact that a respected, skeptical tech journalist came away impressed after a full week of daily use is meaningful signal. It suggests Apple hasn't just dressed up the old Siri with a new coat of paint — something substantive has changed under the hood.
Should You Try the New Siri AI?
If you're an iOS developer or enrolled in Apple's beta program, the new Siri AI is worth exploring right now, with the understanding that bugs and inconsistencies are part of the deal. For everyone else, the public release of iOS 27 is expected later this year, and based on Stern's assessment, it's shaping up to be one of the most compelling reasons to update since Face ID.
The bar for Siri has always been set by user frustration as much as competitor performance. For the first time in a long time, it looks like Apple might finally be clearing it.