Hacker News new | past | comments | ask | show | jobs | submit login

This is a bit out of context. My point is that the article's results should be mapped against hate crime statistics to illuminate a possible rationale for the bias.

I'm not taking a position on whether this bias should inform the results of ChatGPT. I have opinions on the topic, but they are not strongly held.

(Rereading my comment I see why it appears I'm taking a side here. When I said "it makes sense to..." I meant that "[if the intent of OpenAI's moderation was to align moderation strength with statistics then it would] make sense to...")




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: