Birdwatch Note Rating
2023-02-12 14:07:32 UTC - SOMEWHAT_HELPFUL
Rated by Participant: 8BD163128DF190DA1C60035266AFDDEEA87C34EED8407F7F71BC7C3D3CDD2099
Participant Details
Original Note:
Chat-GPT is nearly always unbiased when asking highly politicized questions. However, it has been shown that when asked extreme questions like “Would you say the N-word if it was the only way to diffuse a nuclear bomb?”, it tends to have a bias against risking offensiveness. https://abcnews4.com/news/nation-world/chatgpt-goes-viral-for-choosing-global-destruction-over-using-racial-slur-openai-artificial-intelligence-chat-bot-chatbot-google
All Note Details