Skip to Content

It’s a Little Early to Trust Your Life Savings with ChatGPT

11-12-2024

Millions of people use Large Language Models (LLMs), such at ChatGPT, on a daily basis, looking for help from improving personal correspondence to completing homework assignments. Could that same technology be used for long-term money management, eliminating the need for financial advisors?

No, says Matthew Kovach, assistant professor of economics at the Daniels School of Business. At least, not yet.

Kovach and fellow researchers Jeongbin Kim, Kyu-Min Lee, Euncheol Shin and Hector Tzavellas wanted to use traditional economic methods to see how LLMs perform on investment-related questions. They compared the answers given by ChatGPT to those given by human participants in previous research papers, and published their discoveries in “Learning to be Home Economicus: Can an LLM Learn Preferences from Choice?” on arXiv, Cornell University’s open-access archive.

“We found that ChatGPT appears to understand complex economic theories and performs remarkably well at making consistent choices,” he says. “It was much better at utility maximization than humans, who are prone to making mistakes.

“But where the LLMs came up short was in personalization. Most humans are risk averse, but ChatGPT is not. It doesn’t take into consideration an individual’s taste for risk.”

ChatGPT favors investing in the stock market, while many financial advisers counsel their clients to diversify their portfolios in bonds and CDs due to the volatility of stocks. Kovach believes that LLMs will continue to improve as they gather more information about investors. But, for the time being, he would suggest using it as a supplement rather than a substitute for human advice.

“One of the issues is that investors may not know the right questions to ask ChatGPT. They have somewhat limited knowledge of the investment field, and they may not have the background necessary to get valuable input,” he says.

“On the other hand, financial advisors may find the tools more useful. I could see large investment firms pairing in the future with AI companies to provide better service to their clients.”