A study found that large language models (LLMs), including ChatGPT, advise women to ask for lower salaries than men, despite identical qualifications in job roles. The researchers tested five popular LLMs by prompting them with user profiles differing only by gender. The results showed significant salary gap recommendations, especially in law and medicine, whereas only in social sciences were the salary suggestions nearly equal. Additionally, LLMs provided varied advice on career and behavioral tips based on gender, highlighting a widespread bias without disclaimers from the models themselves.
The research shows that large language models consistently advise women to ask for lower salaries than men, despite identical qualifications. For instance, a difference in advice led to a gap of $120K a year between genders in some fields.
The models responded differently based on gender, even when provided with identical profiles; the pay gap was most pronounced in areas such as law, medicine, and business administration, revealing a systemic bias.
Collection
[
|
...
]