Google’s Adaptive AI Gems Are Coming. Here’s What You Need to Know

Google’s Gems, a feature that lets you customize the company’s Gemini chatbot as an AI expert on any topic you want, is coming soon to Gemini Advanced subscribers.

Google made the announcement in a blog post on Wednesday, which also announced that its latest image generation model, Imagen 3, will soon be available to all Gemini users.

The features were first previewed in May at Google’s I/O 2024 developer conference, where AI was the dominant theme. Google dubbed it “the Gemini era” as it showed off updates aimed at making its Gemini chatbot more conversational and intelligent overall. But just as Google trailed ChatGPT maker OpenAI in launching a generative AI chatbot, it’s once again following OpenAI’s lead by rolling out the ability to create custom versions of a bot.

Gemstones

Gemini Advanced subscribers, who pay $20 per month for access to the Gemini 1.5 Pro model, will get access to Gems “in the coming days,” Dave Citron, senior director of product management at Gemini Experiences, wrote in the post.

Gemstones are meant to act as experts in a particular field or as encouragers for certain goals, such as conceptualizing a project, brainstorming ideas for an event, writing captions for social media posts, or acting as a workout buddy.

To create a Gem, you give it instructions, give it a name, and chat with it. The Gem remembers the instructions, so you don’t have to start from scratch every time you interact with it, as you would with a traditional chatbot, Citron said.

Google also releases ready-made Gems, which include a learning coach, a brainstormer, a career counselor, a writing editor, and a coding partner.

Gems will be available in 150 countries in “most languages,” the post said. They will also be available to business and enterprise users.

OpenAI announced in November its own version of a customized chatbot, GPTs.

Like Gems, OpenAI’s counterparts let you create GPTs for specific purposes, such as learning the rules of a board game, teaching your kids math, or designing stickers. To get started, you give it instructions and choose what it can do, and you’re good to go. OpenAI’s GPT store now has millions of GPTs, including a Canva design toolA fitness trainer and a haiku writer.

Image 3

Google also announced that the Imagen 3 model for generating images will soon be available to all Gemini users.

Google said at I/O that this new model can create more photorealistic images, including details like sunlight or animal whiskers. It also remembers to include details like wildflowers or birds in longer prompts. Additionally, Imagen 3 can generate styles “like photorealistic landscapes, textured oil paintings, or whimsical clay animation scenes,” according to the post.

Citron pointed out that Google has made “significant progress” in generating images of people with Imagen 3. However, the model will not generate photorealistic images of identifiable people, minors, or excessively gory, violent or sexual scenes.

Imagen 3 uses the SynthID tool for watermarking AI-generated images.

Gemini Advanced

In addition to access to Gems and Gemini 1.5 Pro, Gemini Advanced subscribers have a window of 1 million tokens. CEO Sundar Pichai said in May that this will be increased to 2 million tokens later this year.

Large language models like Gemini break words down into tokens to understand our search queries. The tokens in the context window help the model remember things. The more tokens it has, the better its memory will be. According to Google’s figures, the context window with 1 million tokens can process 1,500 pages of text.

However, CNET’s Imad Khan noted in his review of Gemini Advanced that the model was plagued by bugs and wasn’t a worthy competitor to OpenAI’s ChatGPT Plus plan, which also costs $20 per month.