Google Unveils Android AI System, This is Apple's Vision of Itself

Just like last year, before the official Google I/O developer conference, Google held a small event specifically for Android.

At this year's Android Show, Google almost unveiled all of its "visions" for Android and the entire AI product ecosystem for the next year in one go.

Apart from the underlying Android 17, we also saw some unexpected platform updates and the release of several hardware products.

The key point is: Although there aren't many products and it will still take some time for Android 17 to be widely released, we can already see Google's plans for the AI ecosystem over the next five or even ten years.

What's more, tonight's Google is exactly what Apple dreams of becoming after entering the AI era.

Operating System to Intelligent System

Right at the start of the event, Google announced a significant move:

Android will no longer be just an operating system (Operating System), but an intelligent system (Intelligence System).

Image courtesy of Google

In other words, the software and hardware usage model dominated by "manual user operation" is considered a relic of the past by Google.

Starting today, Android, as a cluster of smart features, will become more proactive in intervening in the user's workflow, anticipating and fulfilling your needs.

Image courtesy of Google

This is why Android 17, the specific system version, was mentioned only a few times during the event, while Gemini Intelligence became the system's representative name.

Firstly, as the latest form of Google's Gemini on mobile hardware, Gemini Intelligence significantly enhances its multimodal, cross-environment, and highly integrated operational mode.

For example, as the first input entry point of the system, the default keyboard Gboard has been enhanced with new features.

Based on Gemini Intelligence's multimodal capabilities, the previously underutilized "auto-fill" feature in the keyboard now has access to a broader range of information sources:

Image courtesy of Google

In addition to manually saved passwords, Gboard will also support auto-filling information from photo libraries, addresses mentioned in chats, and schedules written in emails, among others.

More simply put: Gemini Intelligence has gone beyond just helping you remember passwords; it truly acts like an assistant, remembering and suggesting various types of information from different sources.

Another effective way to utilize these multimodal information is through Android's widgets.

In Android 17, Gemini Intelligence will support a feature called "Create My Widget," though it won't be available immediately but aims to launch later this year.

Image courtesy of Google

This new feature mainly allows users to create new widgets on their desktop using a vibe coding-like approach based on their instructions, breaking the limitation that widgets are pre-set by apps.

For instance, instead of using a single-purpose calorie-tracking app, you can tell Gemini: "Create a widget that recommends two high-protein meals per week."

Image courtesy of Google

This way, the desktop widget becomes an entry point for composite tasks, essentially similar to vibe coding an app.

Additionally, users who prefer voice input will benefit from the new Gboard, which supports highly intelligent voice input functionality similar to Typeless, named Rambler.

Compared to the previous method of saying "comma... period..." aloud, Rambler can translate, clean up, and organize a whole segment filled with "ums" and "ahs" into neat text:

Image courtesy of Google

On the other hand, the automatic execution capability of Gemini Intelligence has also been further strengthened.

Last year, at Google I/O and the release event, Google demonstrated giving Gemini commands to automatically order food, call a taxi, book tickets, etc., and the response after its official launch was positive.

In Android 17, with Gemini upgraded to Gemini Intelligence, this "proxy operation" now supports multi-step tasks.

For example, while previously it only supported simple tasks like "book a ticket for me," you can now directly take a picture of a travel brochure in the Gemini dialogue and say, "Find a similar double-person tour package on Ctrip":

Image courtesy of Google

The point isn't about what tasks Gemini can perform, but about its enhanced ability to "do one more thing." Sometimes, it's that extra step that turns AI from "usable" to "useful."

Of course, the update of Android 17 isn't just about Gemini. Google has also optimized many "core Android experiences."

In Android 17, Google has partnered with Meta to support native camera functions such as Ultra HDR, super stabilization, and night video in apps like Facebook and Instagram.

And Google's flat-style emojis, used for many years, have received an update—changing from pure 2D to 2.5D, making them more similar to the emojis used on iOS:

Image courtesy of Google

The QuickShare feature, which is compatible with AirDrop, will also support more manufacturers' devices in Android 17.

In addition to Samsung and Pixel, (international versions) OPPO, Vivo, OnePlus, and Honor will gradually update their firmware to be compatible with AirDrop later this year:

Image courtesy of Google

Google's AI PC

Apart from Gemini Intelligence, Google hasn't forgotten to create custom hardware for these more complex and powerful AI functionalities.

This new hardware is neither a pendant, nor headphones, nor a watchband—it's the once familiar Chromebook.

Yes, in the AI era, Google has renamed its notebook again.

From Pixelbook to Chromebook, and now the latest Googlebook:

Image courtesy of Google

Like Chromebook, Googlebook isn't a specific laptop but a brand that can be used by any third-party manufacturer that meets the standards.

Googlebook is the first hardware product designed specifically for Gemini Intelligence.

Apart from the new Gemini Intelligence features, Googlebook has made revolutionary innovations in the most frequently used basic human-computer interaction level.

To activate Gemini on Googlebook, you don't need to speak, press buttons, or use right-click menus—you just need to "shake the cursor."

Image courtesy of Google

Based on Gemini Intelligence's multimodal capabilities, recommended AI commands can automatically adjust according to the content under the cursor, selected content, and possible actions on the screen.

In an era where AI functions are increasingly dense on computers and laptop keyboards are limited in space, Googlebook's "magic pointer" is undoubtedly one of the most intuitive and elegant solutions.

Additionally, Googlebook solves one of the long-standing issues in ChromeOS: it can run Android apps, something that Google products couldn't do before.

In other words, all the apps on an Android phone can run directly on Googlebook, achieving a level comparable to macOS running iOS software.

The foundation for this is the GKI (Generic Kernel Image) plan mentioned in previous articles by iFeng, which is allowing Android to break free from its mobile constraints and seamlessly integrate into various device forms.

Although Google hasn't mentioned the compatibility of this feature yet, we speculate that based on processor specifications and network status, Googlebook should support both local app running and screen mirroring from a phone.

Image courtesy of Google

According to the event announcements, the initial production partners for Googlebook are familiar names like Acer, ASUS, Dell, HP, and Lenovo, with the first products expected to hit the market this year.

Image courtesy of Google

Apart from laptops, Android 17 also updated some features of the Android Auto infotainment system.

For example, Gemini Intelligence's proxy operations similar to OpenClaw, optimized 3D road views, and more intelligent media playback features.

It also supports displaying custom widgets created on phones on the car infotainment system.

Image courtesy of Google

At the same time, the list of brands natively supporting Android Auto is expanding, with some models even supporting memory of current vehicle information, such as trunk size and dashboard specifications.

This way, when using Gemini Q&A, the infotainment system can provide specific answers, like "Can two 27-inch suitcases fit at the same time?" or "What does that assassin-like warning light mean?"

Unfortunately, this feature won't be available immediately either, with products expected to adopt it later this year.

Overall, this event was just a starter for this year's Google I/O, but the conceptual changes involved are quite fundamental—

Actually, at the beginning of the event, Google pointed out: Effective AI technology should be seamless and invisible, integrating into every layer of software and hardware experience.

Image courtesy of Google

That's exactly what Gemini Intelligence is doing.

Whether it's Pixel phones, Android Auto infotainment systems, or Googlebooks, these hardware devices are merely different manifestations of Gemini intelligence.

Interestingly—what Google did tonight is precisely what Apple has been striving to achieve: a unified intelligent system across iPhones, watches, and Macs, ensuring consistent functionality and experience regardless of the device used, with hardware differentiation only affecting interaction methods, not intelligence levels.

Image courtesy of Apple

Unfortunately, despite Apple's efforts, it hasn't managed to solve the "model" part, leaving its hardware to serve as a platform for others' models.

In the future, intelligent systems (Intelligence Systems) will be more diverse in form but more unified in core.