Google launched a beta feature called Personal Intelligence in its Gemini app on Wednesday, enabling the AI assistant to tailor responses by reasoning across users’ Gmail, Photos, Search, and YouTube history data.
Prior to this update, Gemini could retrieve information from these connected Google apps. The new capability allows it to analyze data across multiple sources simultaneously, delivering proactive results. For instance, it links an email thread to a relevant video watched by the user. Google explains that this process enables Gemini to grasp contextual details without explicit instructions on data locations.
Personal Intelligence activates only after users opt in by connecting their Google apps to Gemini, as it remains off by default. This setup respects user preferences, given concerns about AI access to personal photos and YouTube history. Once enabled, Gemini engages Personal Intelligence selectively, applying it solely when the system deems it beneficial for the query.
Video: Google
Josh Woodward, Vice President of the Gemini app, Google Labs, and AI Studio, detailed the feature in a blog post. He stated, “Personal Intelligence has two core strengths: reasoning across complex sources and retrieving specific details from, say, an email or photo to answer your question.” Woodward added, “It often combines these, working across text, photos and video to provide uniquely tailored answers.” These strengths facilitate responses that integrate diverse data types for precision.
Woodward provided a practical example from his experience at a tire shop. While waiting in line, he could not recall his car’s tire size. Standard AI chatbots might offer general tire size information, but Gemini exceeded this by delivering a customized suggestion. It identified family road trip photos stored in Google Photos and recommended all-weather tires accordingly. In the same interaction, when Woodward forgot his license plate number, Gemini extracted it directly from a photo in his Photos library, demonstrating retrieval from visual content.
Beyond automotive scenarios, Woodward reported Gemini delivering strong recommendations in various domains. These include books, shows, clothes, and travel options. Recently, it assisted with spring break planning. By examining family interests and records of past trips from Gmail emails and Google Photos, Gemini avoided popular tourist areas. Instead, it proposed an overnight train trip paired with particular board games suitable for the journey, tailoring the itinerary to documented preferences.
Google implemented safeguards for sensitive information. Gemini refrains from proactive inferences involving topics like health data. It processes such details only upon direct user request. Separately, Google confirms that Gemini undergoes no direct training on individual Gmail data, preserving user privacy in model development.
