What Siri AI on macOS 27 Is – And How It Was Tested
Siri AI on macOS 27 is Apple’s new conversational assistant that combines on-device Apple Intelligence with online models to answer questions, control apps, and work across your files in a more natural, chat-style interface. To gauge how it compares in real use, ZDNET’s Lance Whitney ran Siri AI through the same style of tasks used to evaluate ChatGPT and Gemini, including open-ended questions, personalized recommendations, file search on a Mac, image analysis, and follow-up conversation. Access is limited to a waitlisted, developer-beta build running on supported Apple Intelligence hardware, so performance reflects a pre-release state rather than a final product. Even so, using identical prompts and workflows across assistants offers a fair AI assistant comparison and exposes where Siri AI already keeps up—and where the gaps with ChatGPT and Gemini are still obvious.
Conversation Quality: More Direct Than ChatGPT and Gemini
In day-to-day chat, Siri AI behaves more like a focused tool than a chat partner. When asked “what’s new,” it skipped small talk and responded with a concise rundown of current news stories, showing Apple’s intent to frame Siri as an efficient assistant instead of a conversational buddy. For factual prompts such as “Why did the Roman Empire fall?”, Siri AI produced a brief spoken explanation with bullet-point causes and clickable web sources, broadly in line with what ChatGPT and Gemini might provide but with less narrative detail. According to ZDNET, the new Siri is “more useful than old Siri but still makes mistakes,” and its conversation flow can feel stilted because follow-up questions sometimes require extra prompting to keep the session going. Compared with the smoother, more expansive dialogue of ChatGPT and Gemini, Siri AI’s conversational edge is still underdeveloped.
Productivity and On-Device Tasks: Missed Opportunities
Where Siri AI macOS 27 should shine is deep integration with system data, but the tests suggest mixed execution. When asked for laptop buying advice with a defined budget and priorities such as keyboard quality and battery life, Siri AI initially surfaced links to articles and social posts instead of forming its own recommendation, something ChatGPT or Gemini typically handle in a single, synthesized reply. Only after a follow-up request did Siri AI summarize the sources and provide an opinion, adding friction to the workflow. File-based commands showed similar limits: searching Photos for all images of the Abraham Lincoln statue returned only three matches, though the library held six, raising questions about reliability for power users. These gaps mean that, despite Apple’s promise of richer macOS 27 AI features, Siri vs ChatGPT performance still tilts in favor of rivals on many productivity tasks.
Vision and Personal Help: Inconsistent Accuracy and Awkward Flow
Visual understanding is another area where ChatGPT and Gemini have set expectations, and Siri AI’s early results trail them. When shown a painting by Toulouse-Lautrec, Siri AI misidentified both the artist and the artwork, then misnamed a second painting even after recognizing its creator. Only on a third attempt, with a well-known Van Gogh, did it deliver a fully correct answer. Conversational personal help was more encouraging: when asked for ideas because a cat named Mr. Giggles sometimes refuses his usual food, Siri AI responded with practical suggestions and then asked a relevant follow-up about wet versus dry food before giving more tailored advice. However, the interaction still felt choppy because the assistant did not always keep the thread naturally open, something ChatGPT and Gemini tend to manage with fewer prompts and smoother context handling.
Beta Status, macOS 27 Golden Gate Beta 2, and What Apple Must Fix
It is important to remember that Siri AI on macOS 27 is available only through a waitlisted developer beta, and Apple has several months left before the expected public release. The macOS 27 Golden Gate Beta 2 update continues refining Apple Intelligence and the surrounding interface, including the new Liquid Glass look and the dedicated Siri AI app entry point, signaling that Apple is still tuning both visuals and underlying behavior for developers. In its current form, the assistant is more capable than the old Siri but still trails ChatGPT and Gemini on accuracy, initiative, and conversational fluidity. To compete in AI assistant comparison tests at launch, Apple must improve result completeness for on-device searches, reduce factual errors in visual recognition, and smooth multi-step dialogue so users do not feel forced to micro-manage every follow-up request.
