Conversation

Notices

1. Embed this notice
   Arvind Narayanan (randomwalker@mastodon.social)'s status on Wednesday, 26-Apr-2023 22:50:44 JST Arvind Narayanan

   • Sayash Kapoor

   People have been posting glaring examples of ChatGPT’s gender bias, like arguing that attorneys can't be pregnant. So @sayashk and I tested ChatGPT on WinoBias, a standard gender bias benchmark. Both GPT-3.5 and GPT-4 are about 3 times as likely to answer incorrectly if the correct answer defies gender stereotypes — despite the benchmark dataset likely being included in the training data. https://aisnakeoil.substack.com/p/quantifying-chatgpts-gender-bias

   • Paul Cantrell repeated this.
   • Embed this notice
     Arvind Narayanan (randomwalker@mastodon.social)'s status on Wednesday, 26-Apr-2023 22:50:47 JST Arvind Narayanan

     in reply to

     OpenAI mitigates ChatGPT’s biases using fine tuning and reinforcement learning. These methods affect only the model’s output, not its implicit biases (the stereotyped correlations that it's learned). Since implicit biases can manifest in countless ways, OpenAI is left playing whack-a-mole, reacting to examples posted on social media.