A study reveals significant disparities in chatbot responses based on race and gender-associated names, impacting salary recommendations and more.
AI chatbots exhibit biases towards Black individuals and women in scenarios such as hiring decisions, purchase advice, and political predictions.
Guardrails meant to limit AI model biases sometimes fail, leading to potential discriminatory outcomes in real-world applications, as noted by Stanford Law School professor Julian Nyarko.
[
add
]
[
|
|
...
]