Google’s AI Bard acts just like a student who hasn’t read their lessons when it comes to generating images upon request!
When it comes to text-based requests, Bard, like its counterpart ChatGPT, behaves very intelligently and creatively; but when it comes to producing images from text, the story goes differently.
When asked if it could create visual content, Google Bard replied, ‘yes, I can create various images for you’, and even claimed it could use different methods to do so.
Initially, this conversation shocked the person asking the question, but it became clear later that Google’s highly vaunted AI could not do such a thing.
The user asked Bard to create an image of an orange cat, and in response, the AI created a string of text instead of an image, describing the orange cat and the position it was supposed to be in. Later, they asked him to create a picture of a pink cat, but he argued that there is no such cat in existence.
According to the user, for the last experiment, he asked Bard to create an image of a girl with earrings, but the result was the same. This time, AI generated a text string and provided a description of the book âGirl with a Pearl Earringâ by Johannes Vermeer along with a description of the background of the photo.
Furthermore, Google Bard was asked to create an image, not just a textual description of the image. AI replied: ‘I apologize that I am still unable to produce your requested images. I am still developing and learning how to create images.’
This is the first time that AI confidently speaks about an unreal thing and mistakes the facts. This means remembering a textual description as an image; an issue that developers call a delusional topic in the world of artificial intelligence. However, challenging AI still remains an entertaining task for users.