
Millions of adults now turn to artificial intelligence chatbots for personal money management, seeking guidance on everything from monthly budgets to stock portfolios. While these digital tools generally push users toward sound financial habits, their recommendations also suffer from rigid rules of thumb and subtle demographic biases, according to a study published by National Bureau of Economic Research (NBER).
Smarter savings, but rigid rules
By analyzing prompts written by a representative sample of adults and simulating their lifetime impact using economic models, researchers evaluated how advice from tools like ChatGPT shapes long-term wealth. Following AI guidance moves consumers closer to standard economic life-cycle theory. For instance, virtually all simulated users built emergency savings buffers exceeding $10,000 by age 30, while portfolio recommendations encouraged broad stock market participation with equity allocations that naturally declined after age 45.
However, the AI advice departed from optimal economic principles in key ways. Chatbots relied heavily on simple heuristics, with nearly a third of recommended saving rates landing on multiples of 10% and over 98% of retirement withdrawals capping at the traditional 4% rule. Furthermore, the models failed to adequately cushion spending after job loss and exhibited portfolio inertia, directing new savings rather than actively rebalancing existing investments as markets shifted.
Read also: Workforce 2026: The gendered impact of the AI automation wave.

Demographics and the gender gap
The research also uncovered significant disparities across demographic groups. Prompts written by men, individuals with high financial literacy, or prior AI experience resulted in lifetime advice that accumulated 4% to 6% more wealth by age 60.
To determine why a gender gap existed in stock recommendations, researchers experimentally assigned male or female labels to identical prompts. They discovered that two-thirds of the equity gap stemmed from differences in what men and women asked, known as demand. However, the remaining one-third was supply-driven: the AI models recommended lower equity shares to female users even when their financial prompts were otherwise identical to those submitted by men.