Google revealed on Wednesday the launch of a new beta feature within the Gemini app. This enhancement allows the AI assistant to customize its outputs by integrating with your Google ecosystem, initially covering Gmail, Photos, Search, and YouTube activity.
While Gemini previously could access data from these applications, it can now analyze information across your various data sources to proactively deliver relevant results. For instance, it can link an email conversation to a video you’ve viewed. Google emphasizes that this functionality enables Gemini to comprehend context without explicit instructions on where to search.
The tech giant clarified that this beta experience, named Personal Intelligence, is disabled by default, giving users the choice to connect their Google applications to Gemini as they see fit. Naturally, not everyone is comfortable with AI reviewing their photos and YouTube history. Should you opt to link your apps, Google states that Gemini will only utilize Personal Intelligence when it determines such integration would be beneficial.
“Personal Intelligence boasts two primary capabilities: analyzing diverse and complex data sources and extracting specific details from, for example, an email or photo to address your query,” explained Josh Woodward, VP, Gemini app, Google Labs, and AI Studio, in a blog post. “It frequently combines these, processing text, images, and videos to generate uniquely personalized answers.”
Woodward provided an illustration involving a time he was at a tire shop and couldn’t recall his car’s tire size. While most AI chatbots can determine a vehicle’s tire size, Woodward noted that Gemini can go further by offering customized recommendations. In his scenario, Gemini suggested all-weather tires after identifying family road trip photos within Google Photos. Woodward also mentioned forgetting his license plate number, which Gemini successfully retrieved from an image in Photos.
“I’ve also been receiving excellent suggestions for books, shows, apparel, and travel,” Woodward wrote. “Just this past week, it proved exceptionally useful for organizing our upcoming spring break. By analyzing our family’s interests and previous trips recorded in Gmail and Photos, it bypassed common tourist traps. Instead, it proposed an overnight train journey and specific board games for us to enjoy along the way.”
Google states that it has implemented safeguards for sensitive topics, meaning Gemini will not proactively make inferences about private data such as health information. However, the company also clarified that Gemini will discuss such data if explicitly prompted by the user.
Furthermore, Gemini does not directly train on your Gmail inbox or Google Photos library. Instead, its training focuses on particular prompts within Gemini and the model’s subsequent responses. In the examples provided, the road trip photos, the license plate image in Photos, and the emails in Gmail are not used to directly train the model; they are merely referenced to formulate a response, according to Google.
Personal Intelligence is currently rolling out to Google AI Pro and AI Ultra subscribers in the U.S. Google intends to extend this feature to additional countries and Gemini’s free tier in the future.
Google offered a selection of sample prompts to try, including “Help me plan my weekend in [city i.e. New York] based on things I like to do,” “Recommend some documentaries based on what I’ve been curious about,” or “Based on my delivery and grocery receipts in Gmail, Search history, and YouTube watch history, recommend 5 YouTube channels that match my cooking style or meal prep vibe.”