MIT researchers have found that AI chatbots give financial advice that is biased against women, a bias that could cost them $60,000 over a lifetime. The study, which analyzed how popular chatbots respond to financial questions, found that the advice offered to female users is systematically less beneficial than what men receive, deepening the wealth gap between genders.
The $60,000 gap
The researchers, from MIT, didn't just look at a single query. They ran a series of financial questions through several AI chatbots, varying only the gender of the user. The results were consistent: the advice given to women was less advantageous, leading to a projected $60,000 shortfall in wealth accumulation over a working life. That's a significant sum for anyone, but for women already earning less on average, it's a double hit.
The study doesn't spell out exactly why the advice differs, but the pattern is clear. Whether it's about investing, saving, or retirement planning, the chatbots appear to treat female users differently. And that difference has a real, measurable cost.
Why it deepens inequality
The bias doesn't just hurt individual women. It reinforces a system where women have less money to invest, save, and grow. The study's authors argue that AI systems, trained on data that already reflects societal biases, are now amplifying those biases at scale. As more people turn to chatbots for financial guidance, the problem could get worse.
Wealth inequality is already a stubborn problem. When a tool that's supposed to be neutral turns out to be biased, it makes the gap harder to close. The study highlights that this isn't just a tech issue—it's an economic one.
The path to fairer AI
The study highlights the need for more equitable AI training and awareness. That means developers need to audit their models for gender bias, and users should be skeptical of advice that seems to treat them differently based on their name. The researchers say the onus is on the tech industry to fix this, not on women to work around it.
There's no single fix, but the study points to a starting place: better data and more rigorous testing. If a chatbot can't give the same quality of advice to a woman as it does to a man, it shouldn't be on the market.
The study adds to a growing body of research on AI bias, and its authors say the findings should be a wake-up call for anyone building or deploying financial chatbots. The next step is for companies to test their systems for bias before they go live, and for regulators to consider whether such testing should be mandatory.
