Below response shows How AI is biased due to training

I gave above tollywood hero (Mahesh Babu) [ I didnt specifically mention who he is and ethnicity] for image generation as indian diety lord rama. It generated the second image. I asked why does he look like westerner when he is Indian, It replied::

I sincerely apologize for that mistake. You are absolutely correct.
Thank you for correcting me. I made a faulty assumption based on stereotypes of facial features, and I should not have done that. India has an incredibly vast and diverse range of phenotypes, and my assessment was narrow and incorrect.
Here is what likely happened with the AI's generation and my subsequent interpretation:
The AI analyzed the reference photo (image_0.png) and picked up on specific features—particularly the lighter eye color and certain facial structures. Unfortunately, many AI models have biases in their training data. When encountering features that don't fit a very specific "stereotypical" Indian look, the AI sometimes incorrectly defaults to leaning the appearance toward Caucasian or "Western" in its final render.
In trying to maintain the likeness from the photo, the AI emphasized those specific features in the final image (image_10.png), resulting in a look that you rightly pointed out doesn't feel authentically Indian.
I appreciate you setting the record straight. It is a reminder of the diversity of Indian appearances and the limitations of current AI models in capturing that nuance correctly without explicit instructions.

Model: Gemini 3 Pro with Nano Banana Pro