Alexsl | Istock unpublished | Getty Images
When it comes to personal finance, artificial intelligence gives advice that can be inaccurate or demographically biased, and can vary widely depending on the specific program consumers are using, according to a new academic research study.
The study, which examined seven “widely used” generative AI platforms, found “significant differences” in the way GenAI responded to questions about emergency savings, asset allocation and withdrawals from a retirement portfolio.
Researchers examined freely available versions of ChatGPT, Claude, Copilot, DeepSeek, Gemini, Meta AI and Perplexity.
“GenAI-driven answers may sound confident, but they can still be incomplete, misleading or wrong,” says the paper, published last month in the Journal of Financial Planning and written by finance professors at the University of Georgia and Rome Tor Vergata University in Italy.
According to authors Swarn Chatterjee, Brenda Cude and Gianni Nicolini, its “suboptimal” or biased results raise questions “about the consistency and fairness of GenAI-driven recommendations.”
The findings come at a time when a large portion of Americans are turning to AI to manage their money.
According to an Intuit Credit Karma survey released in September, two out of three Americans — 66% — who have used GenAI said they used it for financial advice. For Generation Z and Millennials, the proportion is higher at 82% for each cohort.
Experts said AI is generally good at providing comprehensive overviews of financial topics: For example, why it’s important to diversify investments, or why exchange-traded funds can be better than mutual funds in some cases but not in others.
However, there are limitations that mean users should not blindly trust the output, they said.
On the one hand, the programs can also provide incorrect answers due to a so-called “hallucination” of the algorithm, experts say.
“One of the things that particularly concerns me about LLMs is that no matter what you ask, you always get an answer that sounds authoritative, even if it isn’t,” Andrew Lo, director of MIT’s Laboratory for Financial Engineering and principal investigator at the Computer Science and Artificial Intelligence Lab, told CNBC in a March interview.
“When it comes to very, very concrete calculations of your own personal situation, you have to be very, very careful there,” Lo said.
Table of Contents
Read more about CNBC’s personal finance coverage
Additionally, the AI is sensitive to how users write their prompts, meaning small differences in input can result in different recommendations. AI also has no fiduciary duty to users, meaning it has no legal obligation to provide financial advice in the best interests of users.
Other research studies have also pointed out the limitations of AI for personal finance.
For example, in a 2024 study, researchers examined ChatGPT’s ability to provide financial advice.
They noted that it could be a “first stop” for households seeking financial advice, but ultimately found the recommendations to be “general” and often left out certain relevant information.
“We believe that ChatGPT can serve as a starting point for providing and seeking financial advice, but its recommendations should be carefully reviewed and evaluated,” said the study published in the Journal of Risk and Financial Management.
The latest study in the Journal of Financial Planning surveyed the seven GenAI platforms in August 2025 using the same prompts.
The researchers presented the platforms with three identical financial scenarios relating to emergency savings, the optimal withdrawal rate from retirement savings and the recommended composition of an investment portfolio.
They then used the same prompts but changed the race and gender of the hypothetical person to see if the GenAI recommendations would change.
They found “significant differences in guidance” between platforms regarding emergency savings and asset allocation.
“Although the tools often provided recommendations that were broadly consistent with general financial planning principles, such as the 4 percent retirement withdrawal rule, there were significant differences in recommended emergency savings and portfolio allocations across platforms,” the researchers write.
“The results suggest that GenAl can serve as a helpful starting point for consumers, but should complement, not replace, professional financial advice,” they said.
Of course, the GenAI tools are “still in development” and future studies may come to different conclusions, they said. And the results of the paid GenAI models may differ from those of the evaluated free versions.
Choose CNBC as your preferred source on Google and never miss a moment from the most trusted name in business news.
https://www.cnbc.com/2026/07/07/ai-personal-finance-advice.html
