On 9 September, Apple officially showcased the ‘new’ Siri to the world, outside of its beta programme. The Siri overhaul had been a long time coming—Apple had first spoken of it back in June 2024. Siri AI, as it is called now, comes after its closest rival, Google’s Gemini on its Pixel 11 series smartphones, has already delivered a set of features that make an actual difference in real world use.
Taking the two out for a spin brings the age-old, Apple versus Google debate back to the surface, with one fundamental question—which trumps the other? Siri, to be sure, is older—having been first launched as an independent app in 2010 before Apple acquired its maker. Google, on that note, somewhat tanked with its first in-phone assistant—Google Now—which was launched two years later, in 2012.
These days though, Google is one of the world’s foremost AI makers, with its Gemini suite of models and applications ranked at the top of most global AI benchmarks. In fact, on 12 January this year, Apple announced a multi-year contract to lease Google’s foundational AI capabilities, to “power future Apple Intelligence features, including a more personalized Siri coming this year.”
On-phone differences
Siri AI’s three key draws are these: one, it can learn a user’s personal character to recommend results; second, it can ‘see’ both through the camera and what’s on a user’s iPhone display, and finally, it is a standalone app in itself.
Google has all of these three capabilities as well. In May this year at its developer conference, Google launched Gemini Intelligence—which includes ‘Personal Intelligence’ on Pixel phones through Android 17. The latter does what Siri AI promises with ‘Personal Context’: the ability to read a user’s apps to ‘understand’ them better.
Other features new to Siri were launched elsewhere long ago. For example, Visual Intelligence (using Siri to search information via the default camera app) came to Android in 2018, and Google’s Circle to Search and the standalone Gemini app both were launched in early 2024.
In terms of the in-person experience, however, ‘been around for longer’ does not necessarily mean ‘better’. Siri, for instance, is an easier interface to access by simply swiping down on the home screen. The dedicated Siri AI app is present to put all searches under one umbrella, and access them via the tiles-like interface. Moreover, the long-press of the power button is for users to access voice-based Siri services, instead of chat-based text ones.
Personal Context in Siri is a work-in-progress experience that, for now, learns only from Apple apps. For instance, if you use Google’s Gmail app instead of Apple’s official ‘Mail’ app, Siri AI won’t fetch, understand or structure any emails or scheduling-related tasks for you. The same is true if you’re using a third-party calendar app like Google Calendar on your iPhone, instead of Apple’s own calendar app.
The same is true for Google as well. For instance, Personal Intelligence on the Gemini app needs users to select which apps Gemini can access. While this list includes a number of third-party apps, the selection doesn’t include several popular options such as WhatsApp, Notion, Substack and others.
Google’s strength lies in the fact that it owns a slew of apps that even iPhone owners use regularly, such as YouTube, Google Photos, all Work apps (such as Docs and Sheets), and even Google Search. This integration makes the ability for Gemini to learn the personal context of a user more powerful and visibly noticeable.
In a host of other features, Siri AI, as used by Mint on Apple’s iPhone 18 Pro Max, is vastly improved. Its in-camera Siri recognizes birds on the fly accurately, and even returns fast and accurate results for fashion searches. Gemini, used on the Google Pixel 11 Pro Fold for this comparison, matches these use cases equally well.
Google’s party trick
However, for now, Gemini has three clear party tricks that trump Siri AI in versatility and niche use cases. These three include the ‘rambler’ mode, ‘live translate’ for media, and a ‘teleprompter’ mode in the Pixel Camera app.
The ‘rambler’ feature is particularly impressive, recognizing search strings and texts in clear English even for queries that are literally ‘rambled’ by a person in a mix of English and Hindi. It picks up the few keywords that it needs for searches and structures a query automatically, and in our usage, it had an accuracy rate of over 90%.
‘Live translate’, interestingly, also succeeds in translating YouTube streams into a supported local language of choice, without disrupting the rest of the video. For the most part, YouTube has started rolling out this feature natively as well—but its utility is likely underrated so far.
Google’s final showpiece, ‘teleprompter’, is not discussed enough—and could swing many users over to the Android stable (and the Pixel ecosystem solely for the camera app). The prompter is accurate, follows the user’s speech and lip movements closely, and over time, can become a must-have for a video creator.
To sum up, Google’s Gemini on Pixel devices does remain a more superior overall AI experience. Apple, however, has indeed closed the gap on essentials, and once third-party apps are integrated into Personal Context, there’s a good chance that it becomes one of the most nifty, practical on-device AI units in the market.