Google has decided to relaunch the portrait generation feature of its Gemini AI chatbot, which was previously suspended due to historical inaccuracies in the generated images. According to Google, the company will roll out an early access version of this feature to Gemini Advanced, Business, and Enterprise users in the coming days, with support for the English language. The key to this relaunch is the introduction of Imagen 3, Google's latest text-to-image generation AI model. This model has been quietly released in Google's AI testing kitchen and will be fully integrated into the Gemini platform, supporting multiple languages. Imagen 3 possesses powerful image generation capabilities, allowing users to generate various images, from realistic landscapes to richly textured oil paintings, simply by providing a simple description. Previously, Gemini was suspended in February due to the generation of historically inaccurate images. However, the newly introduced Imagen 3 model comes with built-in safeguards. According to Dave Citron, Senior Product Manager at Google, this model outperforms other image generation models in terms of performance. It is worth noting that when relaunching the portrait generation feature, Gemini explicitly prohibits users from creating realistic images of public figures, content involving minors, as well as bloody, violent, and pornographic scenes. Citron stated that while each image generated by Gemini may not be perfect, Google will continue to listen to feedback from early access users and continuously improve this feature. Google plans to expand the AI-generated portrait feature to more users and languages in the future. This relaunch signifies a new step for Google in balancing technological innovation with ethical responsibility.