Can AI Really Help You Manage Your Personal Finances?

A study from Applied Economics

Can AI specifically large language models (LLMs) replace the entire system and give everyone personalized financial guidance for free?

A study tested a wide range of today’s most popular AI models including ChatGPT, Claude, Gemini, and Llama across hundreds of real personal-finance questions that touch nearly every corner of household financial life. The results paint a nuanced picture: AI is already shockingly useful, but also very far from flawless.

This article walks through what the study found and what it means for the future of AI-driven financial advice.

---

The Rise of AI as a Personal Finance Assistant

Large language models have exploded into mainstream awareness thanks to their ability to generate explanations, answer questions, write summaries, and reason across complicated topics. They’re trained on vast quantities of text from books, websites, financial guides, Q&As, corporate filings giving them broad literacy across many domains.

In personal finance, that potential is enormous. Consider the challenges many people face:

• Understanding mortgage structures

• Navigating credit scores, loan terms, and repayment options

• Filing taxes correctly

• Budgeting or planning for retirement

• Evaluating investment choices

Financial advisors have historically filled these gaps, but they charge fees that can run from a few hundred dollars (for robo-advice) to thousands per year for traditional planners. Many households simply go without.

If AI could do even part of this work accurately and consistently, it could dramatically expand financial literacy and access. But the key word is accurately and this is where the study provides valuable clarity.

---

How the Researchers Tested AI Financial Literacy

To test LLMs rigorously, not just with a handful of casual prompts, the researchers used two large, structured datasets:

1. Penn State’s MoneyCounts assessments, covering everything from budgeting to retirement to credit cards to mortgage basics.

2. National Financial Educators Council (NFEC) literacy tests, which include beginner, intermediate, and advanced difficulty levels.

Together, these datasets form one of the most comprehensive financial-literacy question banks available. The models were tested zero-shot meaning they were asked questions without examples or hints to simulate real consumer use.