Software Engineering Institute (SEI) Podcast Series · Members of Technical Staff at the Software Engineering Institute
Using Role-Playing Scenarios to Identify Bias in LLMs
·45 min·4 clips
ChatGPT assigned Hispanic names as ranch hands and non-Hispanic names as astronomers, exposing hidden stereotypes in AI.
As heard by us
A focused look at how name-and-role oddities can surface bias in LLMs.
The episode treats bias in large language models as an engineering problem with real trust consequences, and it does so by grounding the discussion in the work of SEI researchers Katie Robinson and Violet Turi.
Why you'd press play
Listen if you want a practical look at how bias shows up in large language models.
Listen to the show on