Google Gemini has gained a variety of new features, including access to the impressive artificially intelligent image generator Imagen 3 and a new custom chatbot feature called Gems

Imagen 3 was first announced at Google I/O earlier this year and was previously only available through the ImageFX experiment It is more realistic and has improved text rendering on images

Gems was first announced in May It is a bit similar to ChatGPT's GPT, allowing the AI to perform certain actions or play certain characters Gems can be tagged to chat instances on the Gemini web

Both new features will only be available to those paying a $19/month Gemini Advanced subscription through Google One or Gemini for Workspace, but access will be expanded over time

Google Gems is a clever idea It allows users to quickly create custom versions of Gemini for specific purposes For example, one could create one for a specific goal, such as marathon training, with a focus on fitness, health, and nutrition [With Gems,] you can create a team of experts to help you think through a challenging project, develop ideas for your next event, or write the perfect caption for a social media post,” the company explained in a statement And the company explains in a statement: “Your Gem can also memorize a detailed set of instructions to help you save time on tedious, repetitive, or difficult tasks”

This is a great way to help you save time and money

This could include giving details about the class you are taking, the topic you want to study, or even your own writing style

Gem works in most languages and 150 countries, but initially it is only available to Gemini Advanced subscribers or those with business or enterprise accounts

Google DeepMind's impressive Imagen 3 image generation model has been available through the Google Labs service ImageFX and has been tested for about a month It is finally being incorporated into the Gemini chatbot

The new model will be available in English to Gemini Advanced subscribers, but will eventually be available in multiple languages to all users worldwide It has safeguards against inappropriate content and copyright It is not clear how much better this will be than Gemini's Imagen 2, which took Google to task for being overly sensitive to certain topics

All images generated using Gemini's Imagen 3 will be watermarked with SynthID to indicate that they are AI generated if used anywhere They will also only be created in square format

According to Google, Gemini will also be able to generate images of people, but will not be able to depict identifiable individuals, minors, or anything sexual in a photorealistic manner

According to Google: “You control the creative process from start to finish If the first image you see does not meet your expectations, simply tell Gemini what you want to change and a new image will appear”

This is one of the most powerful reasons to incorporate an AI image generator into a chatbot: the AI chat model acts as an intermediary, refining the prompts and sending better instructions to the image generator than a human could convey