Unveiling the Secret Gemini Feature in the Android Search App
byRohitPublished:
Unearthing Root-Level Files Revealing Direct Activation of Gemini AI within the Google Search App on Android
A student and researcher investigating hidden Android features uncovered a buried setting deep within Android's root files. This setting activates Google Gemini directly from Google Search, similar to how it functions on Apple iOS. This discovery raises intriguing questions about its purpose and hints at a potential broader rollout of AI in search speculated to occur in May 2024.
Gemini In Search Rumors
While rumors hint at a potential rollout of AI search by Google, the prospect of Gemini becoming a standard feature raises anticipation within the search community. Here's what to expect:
Gemini represents Google's cutting-edge AI model, boasting advanced training, technology, and features that surpass existing models in numerous ways.
One standout feature of Gemini is its native multimodal training. Unlike previous models that combined separate components for different media types, Gemini is uniquely trained to handle images, text, video, and audio simultaneously. This integrated multimodal training enhances Gemini's ability for complex reasoning tasks, surpassing the capabilities of prior models.
Additionally, Gemini introduces an unprecedented scale of the context window, a critical measure of language model power. The context window, measured in "tokens," signifies the amount of data a model can consider at once to make decisions. Gemini's extended context window sets a new benchmark for language model capabilities.
Exploring Context Window Capacities:
When comparing context windows:
ChatGPT has a maximum context window of 32k.
GPT-4 Turbo boasts a context window of 128k.
Gemini 1.5 Pro stands out with an impressive context window of one million tokens.
To grasp the significance of Gemini's context window, consider this: Gemini can process the entire text of the three Lord of the Rings books or ten hours of videos and answer questions about them. In contrast, OpenAI's top context window of 128k can only handle the 198-page Robinson Crusoe book or approximately 1600 tweets. Gemini's vast context window illustrates its superior capacity for comprehensive data processing and understanding.
Leaked Functionality Resembles iOS Implementation
Unveiling Direct Access to Gemini AI on Android:
A recent discovery reveals that Android devices can access Gemini AI directly from the Google App's search bar, mirroring the functionality available on Apple mobile devices.
The official directions for Apple devices closely resemble the functionality uncovered by the researcher within Android.
Here's how Gemini access is described on iOS:
"On iPhones, you can chat with Gemini in the Google app. With a tap of the Gemini tab, unlock a whole new way to learn, create images, and get help while you’re on the go. Interact with it through text, voice, images, and your camera to get help in new ways."
The researcher who leaked this Gemini functionality in Google search found it concealed within Android. Enabling this feature reveals a toggle in the Google search bar, allowing users to swipe and access Gemini AI just like on iOS.
Activating this feature requires rooting an Android phone, granting access to the operating system's fundamental files.
According to the individual who leaked the information, the toggle requires Gemini to be enabled as the mobile assistant. Additionally, the GMS Flags app must be installed with root permission to toggle Google app features on and off.
Requirements include:
Rooted devices running Android 12+
Latest beta version of the Google App from the Play Store or Apkmirror
Installation of GMS Flags app with root permission (GitHub)
Gemini already available within the Google app
The person who discovered this functionality shared on Twitter:
"Google app for Android to soon get toggle to switch between Gemini and Search [just like on iOS]"
[Include screenshot of new search toggle]
Google Set To Announce Official Rollout Of SGE?
Rumors suggest that Google is gearing up to unveil the official rollout of the Google Search Generative Experience (SGE) at the May 2024 I/O conference, where new search features are typically announced alongside other updates.
Eli Schwartz recently shared insights on LinkedIn regarding the rumored SGE rollout:
"The mentioned date did not come from Google PR; however, it's currently planned for launch internally as of last week. Keep in mind that the timeline could still change, considering it's 53 days away. Over the past year, several launch dates have been missed...
...It's also important to clarify what 'launch' entails.
Currently, the only way to access SGE, unless you're part of the beta experiment, is by opting into labs.
Launching means that they'll showcase SGE to users who haven't opted in, although the scale of this rollout could vary significantly."
It remains uncertain whether this hidden toggle serves as a placeholder for a future version of the Google search app or if it plays a role in enabling the rollout of SGE at a later date.
Nevertheless, this concealed toggle provides a potential clue for those interested in understanding how Google might introduce an AI-driven search interface and whether this toggle is somehow linked to that functionality.
