Exploring Siri AI on Mac: My First Day Experience

I disabled Siri on my Mac years ago and never looked back. I found Apple Intelligence to be pretty unhelpful, so I rarely interacted with it. However, the new Siri AI in macOS 27 Golden Gate has me reconsidering my stance a bit.

I’m still in the early stages of testing Siri AI, having had access to it in the macOS 27 developer beta for just over a day. Since it’s in an early preview phase, there’s plenty of room for enhancements before its official release later this year. I’m not even sure if it’s finished indexing my files and folders on the M5 MacBook Air and M5 Max MacBook Pro I’m using for review. Unlike the iOS 27 developer beta, there’s no “indexing in progress” notification in the settings. When I asked Siri about it, it directed me to a nonexistent button in Settings.

My colleagues have already started testing Siri AI on the iPhone and Apple Watch, and they’ve shared some positive impressions. My experience, however, is a bit more mixed.

When I sit down at my laptop, I don’t really need a voice assistant for quick searches or checking the weather like I would on my phone; I can do that much faster and more accurately with a keyboard and mouse. So, I tried to think of ways Siri AI could actually assist me in my daily tasks on macOS.

I’d love to automate some of the tedious benchmarking I do when reviewing laptops, but while Siri AI can launch apps, it can’t perform actions within them (not that Apple ever claimed it could). I then attempted to create a couple of Shortcuts to see if that could help. This isn’t a Siri AI feature, but it is a new aspect of Apple Intelligence. I asked Shortcuts to run a test in either Geekbench or Cinebench, capture the results in a screenshot, wait a few minutes, and repeat the process two more times. Unfortunately, the automations couldn’t actually execute the tests. Apple Intelligence created a shortcut that opened Geekbench and took screenshots but didn’t run the benchmark, and the Cinebench shortcut included a step that required me to run the test manually. Perhaps if developers continue to expand App Intents, this could work in the future.

If Siri can’t help me run benchmarks, maybe it can at least speed up my data logging. In my usual workflow, I run each benchmark three times, taking screenshots along the way, and later average the results for my spreadsheet. Apple’s WWDC keynote showcased someone using Ask Siri in Spotlight to analyze data in local files, so I tried selecting batches of those screenshots in Finder and asked Siri to calculate the average scores. It worked reasonably well most of the time.

Siri was smart enough to differentiate between single-core CPU scores, multicore CPU scores, and GPU scores, averaging the test results and organizing them into easy-to-read tables. However, it sometimes got confused if I included too many different types of tests, especially when mixing synthetic score results (like Geekbench and PugetBench) with time-based results (like Blender render tests and our 4K video export test). It also occasionally misinterpreted the CPU rankings data visible in Cinebench screenshots. Ideally, I’d want Siri AI to accurately calculate the 15 or so averages from my numerous screenshots all at once, which would save me a lot of time. For now, it can only assist me to a limited extent, and given that it miscalculated a few times by pulling the wrong data, I’m still inclined to handle it all myself.

So far, Siri AI appears to be more effective within Apple’s ecosystem than with non-Apple apps and files on my Mac. When I asked Siri to find my pictures of cats or babies, it pulled results from Apple’s Photos and Messages apps. This might suffice for many users, but not for me. Most of my messaging happens in Signal, and my phone photos are uploaded to Google Photos, not iCloud. Siri also overlooked the thousands of images in my Lightroom Classic catalog, even though those files are stored locally in the Pictures folder and I specifically asked it to access them. It’s possible those files haven’t been indexed yet, but I have no way to confirm.

Currently, I’m getting similar vibes to when I tested Copilot Vision last year. Like Copilot Vision, Siri’s Visual Intelligence allows you to ask questions about what’s on your screen, but it’s limited. I asked Siri to evaluate benchmark results in a Google Sheets spreadsheet, but it couldn’t see all the data if it wasn’t visible on-screen at once. I could get it to view the entire spreadsheet by downloading it as an Excel file and pointing Siri to it in Finder, but when I asked for the laptop with the highest single-core Geekbench score, it provided multicore data instead. Not ideal.

I opened Siri while using Lightroom Classic, on a black-and-white photo from my Ricoh GR IV Monochrome review, and asked how to make it resemble a shot from street photographer Alan Schaller. Siri suggested specific adjustments for exposure, contrast, and more, and tweaking those values yielded a decent result. Unfortunately, when I asked Siri to evaluate the outcome, it overly praised me, claiming I’d achieved an “almost timeless feel,” which is the kind of response Apple says it’s trying to avoid. (I thought we were past this.)

Then I uploaded a classic Garry Winograd photo and asked how to adjust my Lightroom settings to match it; Siri recommended I set the exposure to the value it was already at. So, there were some hits and misses.

It’s still very early for Siri AI, and much can change before the final release. However, it’s already clear that the experience will likely differ significantly between an iPhone, where most data resides within Apple’s apps, and a Mac, where users often navigate various apps and ecosystems that limit Siri’s capabilities. Even so, while it’s faint praise, this is the most useful and helpful Siri has been so far. It feels like Apple is taking its first real steps into AI.