Despite the advancements, the AI tool remains vulnerable to abuse and inaccuracies. To counter this, Google has implemented measures such as embedding watermarks into the pixels of generated images and applying filters to avoid the generation of images of named people. The company continues to invest in new techniques to improve the safety and privacy protections of its models.
Key takeaways:
- Google's conversational AI tool Bard has been upgraded to generate images for free based on user prompts, expanding its training beyond text to video, audio, and photos.
- Despite improvements, Bard's image generation is not 100% accurate and can take around 13 seconds to generate an image.
- Google has implemented safety measures in Bard, including a distinction between visuals created with Bard and original human artwork, and the embedding of watermarks into the pixels of generated images.
- Google is expanding the availability of Gemini Pro in Bard from English to more than 40 languages, allowing it to expand to more than 230 countries and territories.