Google Relaunches Gemini’s People Image Generation After Months-Long Suspension for Quality Improvements

Google has relaunched its Gemini AI’s image generation feature for people after a months-long suspension. The improved tool addresses past issues and is now available to premium users, with expanded features expected soon.

Google Resumes AI Image Generation

Google, a subsidiary of Alphabet, announced on Wednesday that it will resume the capability of generating human faces after a months-long hiatus and that it has upgraded Gemini's AI image-creation algorithm.

According to Reuters, following an assessment of flaws in certain historical pictures produced by the model, Google was forced to temporarily halt its artificial intelligence tool that generates photographs of individuals in February.

Users voiced their disapproval of the problems, which included the fact that the AI model returned historical photographs that were occasionally misleading.

Historical Image Flaws Addressed

The business claimed to have followed "product principles" and conducted tests to identify product flaws, as well as to enhance the product.

Per Yahoo Finance, prior to expanding the model to include additional users and languages, the feature will be made available to premium customers of the Gemini AI chatbot in English.

Imagen 3 Model Upgraded for Accuracy

Though it would not produce photographs of particular individuals, minors, or explicit material, Google claimed to have enhanced the Imagen 3 model to produce better human images.

AI chatbots such as OpenAI's Dall-E, Microsoft's CoPilot, and Grok, which was only recently released by xAI, are among those that are now capable of generating visuals.

According to the search engine giant, subscribers to Gemini Advanced, Business, and Enterprise will soon be able to speak with "Gems" or purpose-built chatbots.

Gemini Subscribers Can Build Custom Chatbots

Users have the ability to build a Gem and provide specific instructions for certain purposes, which saves them time that would otherwise be spent rewriting prompts for use cases that are repetitive.