As Google reintroduces human image generation with its new Imagen 3 model, it aims to tackle past issues with historical inaccuracies and racial depictions.
Google’s new model includes safeguards that prohibit generating identifiable individuals or graphic content, which addresses concerns associated with past image generation practices.
Initial testing of the Imagen 3 model shows significant improvement in avoiding historically inaccurate racial portrayals, addressing the concerns that led to the previous suspension of human image generation.
The ability to generate human images is gradually being rolled out to a subset of users, with particular attention paid to avoiding controversial content.
Collection
[
|
...
]