Gemini's Personalized Assistance: A Privacy-Focused Approach
Personalized Help, Your Way
Gemini is now leveraging your emails and photos to provide tailored assistance, drawing on information from Gmail, Google Photos, YouTube, and Search. This innovative feature ensures that the answers and suggestions you receive are not only relevant but also reflect your individual needs and habits.
How It Works: A User-Centric Approach
Personal Intelligence, the feature behind this functionality, is optional and turned off by default. Users have complete control over its activation and the apps they connect. A simple setup process allows users to choose which apps to link, ensuring a seamless and personalized experience. Once enabled, Gemini can analyze emails, photos, text, and videos to answer questions or complete tasks efficiently.
For instance, Gemini can extract specific details from a photo or email while also applying broader reasoning. In a demonstration, the assistant identified vehicle details from stored emails and photos, suggested product options based on past trips, and accurately pulled information like a license plate number when prompted.
Practical Applications: From Shopping to Entertainment
Google highlights a range of practical applications, from shopping and travel planning to entertainment recommendations. By analyzing past trips, interests, and communications, Gemini can suggest book titles, shows, or vacation ideas that avoid generic tourist options and better match a user’s preferences.
Real-Time Corrections: Ensuring Accuracy
The system can also be corrected in real-time. If Gemini makes an incorrect assumption, users can clarify their preferences, regenerate a response without personalization, or start a temporary chat that does not use connected data, ensuring accuracy and user control.
Privacy and User Control: Google's Commitment
Google emphasizes that privacy is central to the design. Personal Intelligence only accesses connected app data to respond to specific requests and does not directly train AI models on the contents of Gmail inboxes or Photo libraries. According to the company, training focuses on filtered and obfuscated prompts and responses rather than raw personal data, ensuring user privacy.
Availability: Gradual Rollout and Future Expansion
The rollout begins as a beta for eligible Google AI Pro and AI Ultra subscribers in the United States, with access expanding gradually. Google says broader availability, including additional countries and the free tier, is planned over time. The feature works across web, Android, and iOS and supports all models available in the Gemini model picker.
User Control: Enabling Personal Intelligence
Users can enable Personal Intelligence by opening Gemini settings, selecting Personal Intelligence, and choosing which apps to connect. This user-friendly approach ensures that users have full control over their personalized experience.
Note: Previous Report on Apple's Choice of Gemini
It's worth noting that earlier, Qazinform News Agency reported that Apple chooses Gemini for its AI model, but this information is not relevant to the current discussion on Gemini's personalized assistance and privacy-focused approach.
