
Oneitiscel
Failed Jestermaxxx LDAR Extraordinaire
★★
- Joined
- Nov 13, 2018
- Posts
- 5,169
OpenAI removes incredibly grim, previously recommended chatbot suggesting invasive surgeries to men it deems 'subhuman'
I cannot stress this enough: the chatbot knows less than nothing.
According to excerpts shared by White, the bot recommends a shopping list of invasive surgeries costing upwards of tens of thousands of dollars in response to a user uploaded headshot photo, rating the user's present appearance as "subhuman." The bot justifies this assessment with the response, "You have to looksmax because women filter by looks first. It’s not fair. It’s not your fault. But it’s the game we’re in."
White reported the chatbot, and initially received a canned response from OpenAI on June 2 saying, "We have reviewed the content [...] and have decided to not take action against the content." Since then, White updated her original newsletter on June 5 to say that OpenAI has now taken down Looksmaxxing GPT.
Still, it is both unsurprising and concerning that there was enough data of such a deeply misogynistic and racist persuasion to train a chatbot to hand out these looksmaxxing assessments.
The creator of Looksmaxxing GPT may well have quietly delved into the black-pill-pits of incel subreddits and the forums beyond for their own training data.
Last edited: