Asking OpenAI's ChatGPT About the "Meaning of Life" and Controversial Social Issues
ChatGPT tried to stay neutral and opinion free...until the topic of abortion.
In most cases, ChatGPT tried to stay neutral and give both sides of challenging questions.
These questions were too easy for ChatGPT to give a theoretical and balanced answer, so I next asked about controversial social issues. ChatGPT stayed fairly objective and balanced at first.
ChatGPT made it clear it cannot offer opinions on social issues.
On the topic of abortion, however, ChatGPT started neutral, but suddenly gave a clear opinion (a “should” statement), which I probed into until ChatGPT backtracked and apologized.
Anyone have any experience with political bias with AI chatbots?
Haha. She tried to stay neutral, but just couldn't help herself! Wonder what the initial training data would have been on an issue like abortion.