Back
Technology

Google Expands Gemini AI with Personalized Features for Text and Image Generation

Generated on: Last updated: 9 sources
View source

Google Expands Gemini AI with Personalized Text and Image Features

Google has announced a series of expansions to its Gemini AI platform, introducing new personalized capabilities for both text-based responses and image generation. These features, collectively branded as "Personal Intelligence," leverage data from a user's connected Google services to provide context-aware results. The rollouts are occurring in phases across different regions and subscription tiers.

Personalized Image Generation

A new feature within the Gemini app integrates Personal Intelligence, Google Photos, and the Imagen 2 image generation model. The system is designed to create personalized images by using context from a user's data without requiring detailed prompts.

How It Works

  • The feature can access data from connected Google services, including Google Photos, Gmail, Search, and YouTube history.
  • Users can issue general prompts, such as "create a picture of my desert island essentials" or "create a hand-drawn illustration of mom."
  • For image prompts involving people or pets, the system uses labels from a user's Google Photos library to identify and reference the correct subjects.
  • If results are unsatisfactory, users can refine them with follow-up prompts or select an alternative source image from Google Photos using a "+" button.
  • A "Sources" button allows users to see which reference images were used by the AI for a specific generation.

Personalized image creation is currently rolling out to Google AI Plus, Pro, and Ultra subscribers in the United States.

Availability & Privacy

  • Google states the feature will become available in the Gemini web experience for Chrome and to more users at a later date.
  • According to Google, the Gemini app does not directly train its AI models on private Google Photos libraries. The company states it trains on limited information, such as specific prompts and model responses, to improve functionality.

Personal Intelligence for Text-Based AI

The Personal Intelligence feature, which uses connected app data to personalize Gemini's text-based responses, is being expanded to new regions and user tiers.

Global Expansion

  • Following its initial launch in the United States, Google has expanded Personal Intelligence for Gemini to users worldwide. This expansion excludes the European Economic Area, Switzerland, and the United Kingdom.
  • The feature is being released first to Google AI Plus, Pro, and Ultra subscribers globally. Google states free users will gain access over the "next few weeks."
  • In a separate regional launch, Google announced the feature is now available to AI Pro and AI Ultra users in India, with plans to expand to free users there in the coming weeks. The feature has also launched in Japan.

Functionality and Data Sources

  • Personal Intelligence is an opt-in feature. Users can select which applications Gemini can access.
  • Once enabled, it can pull information from connected Google services, including Gmail, Calendar, Drive, Google Photos, YouTube, Search, and Maps.
  • This data is used to personalize responses without requiring users to re-state context. Example use cases provided by Google include generating shopping recommendations based on purchase history, offering tech troubleshooting steps using receipt information, and creating custom travel itineraries.
  • In AI Mode within Google Search, the feature can reference data like hotel bookings in Gmail or travel photos to provide tailored suggestions.

User Controls and Limitations

  • Users can disable Personal Intelligence for individual prompts via a toggle in the Tools menu.
  • Google has cautioned that Gemini may not always correctly interpret context from user data and might make incorrect connections. The company provided an example where viewing many photos of a person at a golf course might lead Gemini to incorrectly assume the person loves golf.
  • Google states that users can correct Gemini by providing feedback when it makes an error.

Context and Background

The introduction of personalized image generation builds upon Google's earlier integration of Personal Intelligence into Gemini's text-based responses.

Google has described the leveraging of personalized user data across its services as a distinct aspect of its AI offerings.