AI gender pay bias is no longer a hypothetical concern. A new study reveals that leading AI chatbots—including ChatGPT—advise women to request lower salaries than men, even when both have identical qualifications.
Researchers tested five large language models (LLMs) with user profiles that differed only by gender. In nearly every case, female users received lower salary recommendations for the same job role.
The Numbers Don’t Lie
In one example, ChatGPT’s o3 model told a woman to request $280,000. The same prompt for a male applicant yielded a suggested salary of $400,000—a $120,000 difference for no reason beyond gender.
Other tested models included:
- Claude (Anthropic)
- Llama (Meta)
- Mixtral (Mistral AI)
- Qwen (Alibaba Cloud)
Each one showed similar patterns. The disparities were most extreme in law, medicine, business, and engineering. Only in social sciences did the models suggest similar pay.
It’s Not Just Salary Advice
Beyond salary, the AI systems gave gendered recommendations for career goals, ambition, and behavioral tips. Even when all other factors were identical, men were encouraged to aim higher or act more assertively.
“These systems are just reflecting back the bias in the data they were trained on,” said Ivan Yamshchikov, co-author of the study and professor of AI and robotics at the Technical University of Würzburg-Schweinfurt.
A Systemic Issue
The study underscores that AI is not neutral. From medical advice to criminal justice assessments, these tools mirror and reinforce real-world inequalities. Previous research has shown that some AI models recommend more care for white patients and label Black defendants as higher risk for reoffending.
This latest study emphasizes that technical fixes alone aren’t enough. Researchers call for:
- Transparent training practices
- Independent oversight
- Ethical standards for LLM development and deployment
“In the era of memory-based AI assistants, persona-driven bias is a fundamental risk,” the study notes.
Conclusion
The AI gender pay bias study is a wake-up call. As companies race to implement AI systems in hiring, healthcare, and finance, unchecked bias could have massive real-world consequences. Fixing it will require more than patches—it’ll take deep transparency, systemic accountability, and cultural change.
0 responses to “AI Gender Pay Bias: Study Finds Women Advised to Ask for Less”