Google takes another shot at AI-generated human images after February failure

Google’s latest AI model will allow people to generate images of human forms in the coming days. The company also plans to incorporate the latest version of Imagen 3, which it claims comes with upgrades to generate accurate images. The feature will be available in the English version for Gemini Advanced users

In the latest blog post by Google, a senior director in charge of the Gemini product, Dave Citron shares details about the “improved” Gemini AI bot. Citron talks about incorporating Imagen 3 as Gemini’s latest image generation model. The model is set to produce images of people with some restrictions.

Some of the restrictions expected are that the photos will not be photorealistic, and it will not allow the generation of images of known faces, minors, or gory scenes of any sort. 

The human image generation feature will only process English prompts and will be available to users who have subscribed to Gemini Advanced,

Google’s Gemini AI faced criticism in February

America's Founding Fathers, Vikings, and the Pope according to Google AI: pic.twitter.com/lw4aIKLwkp

— End Wokeness (@EndWokeness) February 21, 2024

Back in February Google’s chatbot caused a stir on social media by generating historically inaccurate images of people. Also referred to as the “woke AI” Google faced a backlash from critics for not testing the product’s accuracy.  The chatbot made errors in depicting the race of German soldiers and America’s founding fathers. 

In response to the backlash, Google’s co-founder, Sergey Brin, also admitted the inadequacy of the product testing process. As a result, the tool was discontinued until recently. 

In a recent blog post, Dave Citron reflects on the events of February. In his statement, he shares the possibility of errors with Gemini like any AI tool. He also stated that the feedback of early users will be acted upon. According to the post, “Of course, as with any generative AI tool, not every image Gemini creates will be perfect, but we’ll continue to listen to feedback from early users as we keep improving.” 

Citron also shared in the blog post that the company design principle promotes creative liberty throughout the process. Hence users can make changes by giving instructions to the chatbot in case of any error in the image.