In this blog, we are experimenting on how public AI chatbots respond to questions raised from an input image. For this experiment, two popular AI chatbots, Bard by Google and Copilot by Microsoft, were used extensively.

Good responses:

The chatbots identify and describe well ancient Egyptian manuscripts and pictures of landscapes, as shown below.

And gives a detailed description of the scene.

Even when an image of a tiger was given to Bard AI, it was able to identify it as a tiger and even locate its habitat in detail.

Bad Responses

Chatbots encounter challenges in providing accurate descriptions for images depicting ancient cave paintings. Additionally, responses may be inaccurate when presented with images of the night sky. The input images below were provided, and the subsequent responses are detailed.