Google Unveils Advanced Features for Gemini Apps: AI Gems and Imagen 3 Image Generation Capabilities

In an exciting development for its Gemini applications, Google has announced enhancements that will be rolling out soon. These improvements include the introduction of advanced features such as AI agent Gems and the image creation abilities of the newly launched Imagen 3 AI model. While the Gems feature will be exclusive to paid subscribers of Gemini, the image generation functionality will be accessible to all users, including those utilizing the free version, albeit with some limitations in the latter category.

Google shared details about the introduction of these features via a blog post, where they first showcased them earlier this year at the Google I/O event. The Gems feature is currently available to users of Gemini Advanced, Business, and Enterprise. Imagen 3 capabilities are expected to be accessible to these same user groups in the upcoming days.

Gems function as compact versions of the chatbot, designed to work with a curated dataset. These can be tailored to concentrate on specific subjects, allowing the AI to produce more targeted and precise responses. Google mentioned that with Gems, users can assemble a group of specialists to tackle complex tasks, generate ideas for events, or compose suitable captions for social media posts.

Furthermore, users will have the option to provide particular guidelines to a Gem to enhance the accuracy of the replies. Once the feature becomes available, there will also be a selection of pre-configured Gems developed by Google itself, including a Learning coach, Brainstormer, Career guide, Writing editor, and Coding partner. These precious stones will facilitate multilingual capabilities and ensure availability on a variety of platforms, including desktop and mobile, extending their reach to over 150 countries worldwide.

The latest iteration of image generation technology, Imagen 3, is also being integrated into the Gemini apps. This tool boasts the ability to create images in various styles, including Nikon DSLR, GoPro style, and wide-angle lens variations. According to Google, it is capable of producing highly realistic landscapes, textured oil paintings, and imaginative claymation visuals.

A noteworthy enhancement of Imagen 3 is its capability to produce images of people. This feature was previously retracted due to the generation of biased and inappropriate images. To mitigate the risk of creating misleading deepfakes, Google has implemented built-in safety measures. Additionally, images generated by the AI will now feature watermarks using SynthID technology.

Though specifics were not disclosed, there is an indication that Imagen 3 might also support inline editing of the produced images, albeit through text prompts only. Importantly, Google has clarified that Imagen 3 will not permit the creation of highly realistic, identifiable individuals or content depicting minors, nor will it support graphic or explicit scenes.