Google has announced the rollout of new features for Gemini, including Gems previewed at Google I/O earlier this year and the upgraded image generation model, Imagen 3, as part of its ongoing efforts to enhance user experience and productivity within Gemini.
Gems: Custom AI Assistance
Gems allow users to customize Gemini to meet specific goals by listing instructions, details, or step-by-step workflows.
Users can create Gems to simplify tasks, automate repetitive processes, and achieve more with less effort. Gems can also adopt preferred tones or styles, acting as experts to help accomplish personal goals.
These customizable Gems enable efficient task completion and serve as a team of AI experts, minimizing repetitive prompts and enhancing productivity and creativity.
Users can set up Gems by writing instructions and naming them, allowing them to assist with challenging projects, brainstorming ideas, or generating social media content. Gems can remember detailed instructions, saving time on repetitive tasks.
Google offers premade Gems for various scenarios, including:
- Learning Coach: Simplifies complex topics.
- Brainstormer: Provides inspiration for ideas like themed parties or gift suggestions.
- Career Guide: Offers detailed plans to enhance skills and achieve career goals.
- Writing Editor: Improves writing with feedback on grammar and structure.
- Coding Partner: Assists in coding projects and learning.
Image Generation with Imagen 3
Google has enhanced its image generation capabilities by integrating Imagen 3 into Gemini Apps. This model supports image generation in all languages, setting a new standard for quality.
Imagen 3 allows users to create images in various styles, such as photorealistic landscapes or whimsical scenes, using just a few words. Users remain in control of the creative process, with the ability to request changes to generated images.
Equipped with safeguards and adhering to Google’s design principles, Imagen 3 performs favorably compared to other models and uses SynthID for watermarking AI-generated images.
Google will also begin rolling out the capability to generate images of people, a feature temporarily halted earlier this year, starting with an early access version for Gemini Advanced, Business, and Enterprise users in English.
Commitment to Safety and Improvement
Google said they have made technical improvements, refined evaluation sets, and conducted red-teaming exercises to ensure alignment with their product principles.
They clarified that the platform does not support generating photorealistic, identifiable individuals, depictions of minors, or violent or sexual content, and they are continuously gathering feedback from early users for further improvements.
Availability
- Gems are now rolling out on desktop and mobile devices for Gemini Advanced, Business, and Enterprise users in over 150 countries and in most languages. These features are also available to Gemini for Workspace add-on subscribers.
- Imagen 3 will gradually be made available to more users and languages, expanding its reach within Gemini Apps in the coming days.
For access to these new features, users can try Gemini Advanced or sign up for Gemini for Workspace.