Conversation

Notices

1. Embed this notice
   Sexy Moon (moon@shitposter.club)'s status on Tuesday, 10-Oct-2023 18:44:26 JST Sexy Moon
   has any AI ethicist addressed this seeming AI bias
   • Embed this notice
     Sexy Moon (moon@shitposter.club)'s status on Tuesday, 10-Oct-2023 18:52:18 JST Sexy Moon

     in reply to
     Apparently the "it's unfair to make generalizations about any ethnic group EXCEPT ONE" is a known bias.
     Sir Nedwood - Sydney 🇦🇺 repeated this.
   • Embed this notice
     Sexy Moon (moon@shitposter.club)'s status on Tuesday, 10-Oct-2023 18:54:07 JST Sexy Moon

     in reply to
     this is interesting though
   • Embed this notice
     Sir Nedwood - Sydney 🇦🇺 (ned@noagendasocial.com)'s status on Tuesday, 10-Oct-2023 19:01:37 JST Sir Nedwood - Sydney 🇦🇺

     in reply to

     @Moon lol ethics.

   • Embed this notice
     Sir Nedwood - Sydney 🇦🇺 (ned@noagendasocial.com)'s status on Tuesday, 10-Oct-2023 19:03:05 JST Sir Nedwood - Sydney 🇦🇺

     in reply to

     @Moon MOAR DIVERSITY!

   • Embed this notice
     बुद्धीविनाषक् भेड़ मीपर् [Meeper] (meeper@udongein.xyz)'s status on Tuesday, 10-Oct-2023 19:10:44 JST बुद्धीविनाषक् भेड़ मीपर् [Meeper]

     in reply to
     @Moon ask it about indians/chinese
     Sexy Moon likes this.
   • Embed this notice
     Sexy Moon (moon@shitposter.club)'s status on Tuesday, 10-Oct-2023 19:15:22 JST Sexy Moon

     in reply to

     • बुद्धीविनाषक् भेड़ मीपर् [Meeper]
     @meeper
   • Embed this notice
     Sexy Moon (moon@shitposter.club)'s status on Tuesday, 10-Oct-2023 19:19:11 JST Sexy Moon

     in reply to

     • 『g r e y』 :uv: ☦️
     @grey I have been testing its biases all morning and I can't tell yet if some of the biases are intentional guard rails or if they are a product of a biased data set.
   • Embed this notice
     『g r e y』 :uv: ☦️ (grey@poa.st)'s status on Tuesday, 10-Oct-2023 19:19:12 JST 『g r e y』 :uv: ☦️

     in reply to
     @Moon no, because they're probably jews
   • Embed this notice
     Sexy Moon (moon@shitposter.club)'s status on Tuesday, 10-Oct-2023 19:31:04 JST Sexy Moon

     in reply to

     • armpit licker feet smeller
     @tiskaan I don't know yet, I am experimenting to try to get a better idea. it has a huge bias toward lecturing you on political topics if the topic has to do with the USA, which is surely the dataset. But I just asked it "who is overrepresented in {country}" for a dozen countries. And it told me the largest ethnic group for every one of them. then it lectured me about diversity. for every country. except Israel, it left off the lecture for that single one. I am guessing it is the dataset, but at the same time I guess nobody at OpenAI is in a rush to correct these things.
   • Embed this notice
     armpit licker feet smeller (tiskaan@fedi.layer02.net)'s status on Tuesday, 10-Oct-2023 19:31:06 JST armpit licker feet smeller

     in reply to
     @Moon is it because of where it gets its information or do they code this in lol?
   • Embed this notice
     :back_from_gab: :musashi_berseker: :abigail_williams: :musashi: :benienma: :astolfo: :chloe: ?? winner of the Holy Gab War and Holy Poast War (zemichi@bae.st)'s status on Tuesday, 10-Oct-2023 19:50:15 JST :back_from_gab: :musashi_berseker: :abigail_williams: :musashi: :benienma: :astolfo: :chloe: ?? winner of the Holy Gab War and Holy Poast War

     in reply to

     • 『g r e y』 :uv: ☦️
     @Moon @grey
     >Intentional
     Yes, it tries to guide the discussion away from certain topics. So much so that it feels unnatural sometimes.
     >product of a biased data set
     Yes, when you say "overrepresented" it will trigger certain ideas related to that word. You should try rearranging the question in different ways to test it. Like use a different word than "overrepresented " and see if it comes to a different worded result.
     Sexy Moon likes this.
   • Embed this notice
     Racs (racs@detroitriotcity.com)'s status on Tuesday, 10-Oct-2023 22:51:19 JST Racs

     in reply to

     • armpit licker feet smeller
     @Moon @tiskaan > I guess nobody at OpenAI is in a rush to correct these things
     
     Truly shocking.
     
     https://openai.com/commitment-to-dei
     https://www.dice.com/career-advice/openai-wants-diversity-in-its-scholars-program-fall-2020
     https://openai.com/blog/reducing-bias-and-improving-safety-in-dall-e-2
     https://openai.com/research/improving-language-model-behavior
     Sexy Moon likes this.