You can now generate images with Bard, Google's AI chatbot.
Today, the company announced Bard's new text-to-image capability that's powered by Imagen 2, Google's latest diffusion model. With this release, Google also shared that any image created with Imagen 2 — on Bard or Google's other generative AI tools — will have digital watermark called SynthID.
Previously, you could upload images to Bard and ask it to recognize or help you with something related to that image. But asking Bard to generate an AI image for you is new.
Google has already launched AI image generation in other products like SGE (Search Generative Experience), which is still in Labs (its testing ground), and Duet AI. Imagen 2 in Bard joins the likes of other image-generating features in chatbots like ChatGPT and Microsoft Copilot.
To ensure safe and responsible creations of AI images, "there's a clear distinction between visuals created with Bard and original human artwork," the Google announcement said. It also bans the generation of "named people" in an effort to avoid the creation of deepfakes like the recent example of Taylor Swift.
Bard's text-to-image generator is currently available in English. Other releases coming to Bard today is the availability of Gemini Pro in all supported languages and countries.
Topics Artificial Intelligence Google
