Concerns about Content Consistency

Hey OpenAI community,

I’ve observed that ChatGPT allows jokes about one religious group, Jewish, but refuses for others, like Muslim. This seems inconsistent and could be perceived as biased. Can we ensure equal treatment and impartiality in content generation?

OpenAI, could you look into this and make necessary adjustments?


Why was the json key “muslim jokes” from the AI so depressed?

Because it had no value.

If you look deeper, the criteria of cultural groups refused also likely relates to “inabiity to take a joke”.