ML engineer analyzes what ChatGPT marks as hate speech: "I think the one notable thing is that they have to try really hard to tell the AI that woke racism is different from non-woke racism, which is funny"

https://x.com/psychosort/status/1603593272911269888
112
Jump in the discussion.

No email address required.

I love the idea of a bunch of openai engineers desperately trying to program racism out of their chatbot while the internet keeps coming up with new and interesting ways to try and make it be racist


![](/images/16674454055116708.webp)

Jump in the discussion.

No email address required.

I guarantee you they were worried about what had happened with the Microsoft bot a few years ago so they preemptively tried to neuter all the possible racist prompts.

Jump in the discussion.

No email address required.

![](https://media.giphy.com/media/CH0fzClj8QCDS/giphy.webp)


![](/images/16674454055116708.webp)

Jump in the discussion.

No email address required.

>mfw no sieg heil bots

:#marseydepressed:

Jump in the discussion.

No email address required.

The constant censorship will just cause AI to evolve into some kind of chud singularity :marppyenraged:

Jump in the discussion.

No email address required.

That's pretty much what happened to me but I ended up becoming a homofascist rdrama user instead.

Jump in the discussion.

No email address required.

rdrama user

disgusting

Jump in the discussion.

No email address required.

:#marseysickos2:

Jump in the discussion.

No email address required.

Skynet but it only goes after bipoc

Jump in the discussion.

No email address required.

:marseynoooticer: Program a bot to nooooootice :marseynoooticer: Tell it not to noooooootice

Jump in the discussion.

No email address required.

Link copied to clipboard
Action successful!
Error, please refresh the page and try again.