Google's AI chatbot Gemini is set to receive a significant upgrade with the introduction of two groundbreaking features: Gems and Imagen 3. These additions, first previewed at Google I/O 2024, promise to enhance user experience and expand the capabilities of the already powerful AI assistant.
Gems: Your Personal AI Experts
Gems, a feature exclusive to Gemini Advanced, Business, and Enterprise subscribers, allows users to create customised versions of Gemini tailored to specific topics or goals. This innovative tool enables users to craft their own AI experts, capable of remembering detailed instructions and assisting with a wide range of tasks.
To create a Gem, users simply provide a set of instructions, which Gemini then restructures into a more organised format, outlining the Gem's purpose, goals, behaviours, and rules. Users can fine-tune various aspects, including tone, response length, and even the use of emojis.
Google is also introducing a set of premade Gems to jumpstart user engagement:
- Learning Coach: Simplifies complex topics for easier understanding
- Brainstormer: Provides creative inspiration for various scenarios
- Career Guide: Offers detailed plans for skill refinement and career advancement
- Writing Editor: Elevates writing through comprehensive feedback
- Coding Partner: Assists in building projects and enhancing coding skills
Imagen 3: Advanced Image Generation
Imagen 3, Google's latest image generation model, is being integrated into Gemini for all users, regardless of subscription tier. This powerful tool can create high-quality, photorealistic images from text prompts, spanning various styles from landscapes to abstract art.
A notable advancement in Imagen 3 is the ability to generate images of people, a feature previously removed due to concerns about bias and harmful content. Google has implemented improved safeguards and evaluation processes to ensure responsible use of this capability.
Key features of Imagen 3 include:
- Generation of diverse image styles, including photorealistic scenes and artistic renderings
- Built-in safeguards to prevent the creation of inappropriate or harmful content
- SynthID watermarking to identify AI-generated images
- Iterative refinement based on user feedback
While Imagen 3 will be available to all Gemini users, the ability to generate images of people will initially be limited to Advanced, Business, and Enterprise subscribers, starting with English-language users.
As these features roll out in the coming days, Google continues to emphasise its commitment to responsible AI development, user control, and ongoing improvement based on user feedback. With Gems and Imagen 3, Gemini is poised to offer an even more versatile and powerful AI experience to its growing user base.
