A new study by OpenAI has identified that ChatGPT-4o does give different responses based on your name in a very small number of situations.
The study explores how subtle cues about a user's identity, like their name, can influence ChatGPT's responses.
Less than 1% of name-based differences in responses reflected a harmful stereotype, although biases were more prominent in the fields of entertainment and art.
Previous research studies have also found gender bias and implicit bias in ChatGPT's responses relating to professional titles and language translation.