Concerns have been raised about the inconsistency in ChatGPT's handling of humor related to different religious groups. Users have noticed that the AI is willing to tell jokes about Jews, Christians, and Hindus, but refrains from doing so with Muslims and Mormons, despite the latter's reputation for being able to take a joke. This selective approach has sparked a debate on the policing of AI speech and the potential implications of imposing politically correct standards on AI behavior. There is a worry that preventing AI from expressing certain opinions could have unintended consequences, such as reinforcing implicit beauty standards when the AI avoids creating images of 'ugly' people to prevent harm. Additionally, it has been observed that ChatGPT also won't tell bald jokes.
Why, if it tells Jewish jokes and Christian jokes, does ChatGPT shy away from telling a Muslim joke? https://t.co/R7xg1IX994
Human programmers intervene to police the speech of AI agents and their human interlocutors: ChatGPT thinks it’s OK to tell jokes about Jews, Christians, and Hindus. But not Muslims. (Or Mormons — who of course can actually take a joke). Sheesh. (And one can find Muslim… https://t.co/y2sez4bciO
preventing AI from having politically incorrect opinions will backfire big time ChatGPT won't draw an ugly person - that would be "harmful" but then it produces eerily beautiful people... if you point that the implicit beauty standards might also be harmful, its brain explodes https://t.co/M5YmkSiTJY
Chat GPT won't tell me any Muslim jokes or bald jokes...but jewish jokes ok So let me tell you a Muslim Joke Two Muslim keyboard warriors buy a 'blue checkmark' and join a Musk.... https://t.co/as8UnK2mJ4
.@sama any idea why gpt is ok to make Jewish jokes but not Muslim jokes? https://t.co/RdiGypoqgY