Artificial Intelligence

ChatGPT: A Flawed Financial Planner That Could Cost You A $500,000 Dream Home

In recent months, artificial intelligence (AI) softwares are becoming more commonly used by everyday consumers as tools like ChatGPT have begun to offer free, widespread access to anyone with a computer. 

In seconds, users can create a free account and ask the software for help with virtually anything from “help me write a business plan,” to “what should I do for my 10-year-old’s birthday party.” It seems as though the stream of knowledge an AI software can provide is endless, but what are the limitations of this technology, and when it comes to your personal finances are your money-related questions better left to the (human) pros. 

What exactly is ChatGPT? 

ChatGPT is a chatbot that was first launched in November of 2022 by OpenAI, an AI research and deployment company. The software can answer virtually any question in a conversational way, keep track of what a user said earlier on in the conversation and answer follow-up questions. 

“Modern AI derives its core language understanding from public domain datasets like Wikipedia and CommonCrawl. The AI algorithms purposefully corrupt the source data so that the AI can learn how to repair the corruption,” says Kyle McIntyre, Head of AI Engineering at Quiq, a conversational AI and business messaging platform. “In the process of learning how to repair the corruption, the AI learns a great deal about the underlying semantics of the language and can regurgitate and adapt much of what it has learned, assuming a sufficiently powerful AI model.” 

ChatGPT clearly lists its capabilities as: 

  • Remembers what user said earlier in the conversation
  • Allows user to provide follow-up corrections
  • Trained to decline inappropriate requests

Its limitations include: 

  • May occasionally generate incorrect information
  • May occasionally produce harmful instructions or biased content
  • Limited knowledge of world and events after 2021

The state of artificial intelligence as we know it

Over the past few years, the adoption of AI by several large companies has served as a catalyst for expansion and greater financial returns. From the sheer speed of this technology to its long list of capabilities, the concept of AI is an appealing one for company heads looking to speed up production at a lower cost. 

While there was some dropoff in AI adoption during the pandemic, a recent report by McKinsey shows that in 2017, 20% of organizations reported adopting AI in at least one business area. Today, that percentage has more than doubled to 50%. Some of the most popular use cases for businesses range from service operation, product and/or service development, marketing and sales development, and more. 

But for everyday consumers, AI plays a slightly different role. 

Research by shows that 77% of people use a service or device that’s powered by AI. AI is present in everything from your streaming suggestions, to the voice command you use on your mobile or smart devices, your preferred ridesharing app, and even your robot vacuum. 

It’s no surprise that it would infiltrate your personal finances as well. Many major financial institutions have adopted AI to help process loan applications, detect fraud, and manage your investments with the help of a robo advisor

Putting ChatGPT to the financial test  

I wanted to know if ChatGPT is capable of answering some of our most important financial questions and creating a plan for hitting those goals that was tailored to each user’s unique financial stats. To test whether or not the advice provided could align with what a certified financial planner might advise, we gave ChatGPT and a human CFP the following scenario (the figures in this fact pattern are hypothetical for the sake of this experiment): 

  • I’m interested in purchasing a $500,000 home within the next 5 years
  • I live in a single-income household and earn $100,000 per year, before taxes
  • I have the following debts:
    • $30,000 in Federal student loan debt at a 5% interest rate
    • $2,500 in credit card debt at a 22% APR
  • Some of my recurring monthly expenses include:
    • Car payment and insurance: $350
    • Rent and household expenses: $3,000
    • Retirement contributions: 5% of my monthly income with a 5% employer match
  • My savings account balance is $3,000
  • My brokerage account balance is $2,000, invested in various mutual funds and ETFs

ChatGPT took just a few seconds to generate the following response: 

“Based on the information provided, it’s important to focus on reducing high-interest debt first before saving for a home purchase. Here are some steps you can take:

Prioritize paying off your credit card debt as soon as possible. Credit card interest rates are typically much higher than other types of debt, and carrying a balance can quickly add up over time. Try to pay more than the minimum payment each month to make faster progress on paying off this debt.” 

Coincidentally, this was also the first course of action that Matt Mondoux, CFA®, CMT, CFP®  and senior financial advisor at Blue Chip Partners recommended as well. “I would say over the course of the next month or two, just eliminating credit card debt, I think I would certainly speak to the inefficiency of that debt.” 

The next course of action was to tackle the $30,000 student loan debt. Chat GPT recommended the following: 

“Consider refinancing or consolidating your federal student loan debt to a lower interest rate. This can help you save money on interest over time and reduce your monthly payments.” 

Mondoux also recommended tackling the $30,000 student loan debt and making it a goal to repay that amount by the end of the five-year deadline. He thoroughly explained what my minimum payment would be given today’s interest rates, how much I would need to set aside each month to eliminate that debt before assuming new debt in the form of a mortgage, and the benefit of doing so in a timely manner. 

The next step to homeownership according to ChatGPT—building an emergency fund. 

Build up an emergency fund with at least 3-6 months of living expenses. This will help you cover unexpected expenses without going into debt and ensure that you’re financially stable before taking on a mortgage.” 

Mondoux’s advice differed here, accounting for competing priorities and suggesting a two-fold approach: saving for a down payment and building a hefty emergency fund at the same time. When preparing to buy a home, your down payment will play a role in determining what your monthly mortgage payment looks like and your repayment timeline. Working toward a sizable downpayment and an emergency fund for any hiccups that could crop up along the way gives this homebuyer the greatest amount of financial flexibility as they prepare to make one of the most expensive purchases in their lifetime. 

“Taking all of those facts together, here’s the money coming in in the form of salary, here’s the money going out, I look for cash inflows and cash outflows,” says Mondoux. “After taxes, I determined that this individual would have roughly $24,000 of cash coming in and after every other expense, they would have an excess cash flow situation.” 

With a goal of a 20% down payment, Mondoux would then suggest ways to feasibly save $20,000 per year to hit that down payment goal. He also accounted for some of the unexpected expenses that could arise, say a medical emergency, and how to be proactive about them so that they don’t throw your original homeownership goal off track. 

ChatGPT then suggested saving for a down payment and cutting down on monthly expenses to reduce debt. 

“Once you have paid off your credit card debt and have a sufficient emergency fund, start saving for your home purchase. It’s recommended to save at least 20% of the home price for a down payment to avoid paying private mortgage insurance (PMI). Consider reducing your monthly expenses, such as finding a more affordable place to live or cutting back on discretionary spending, to free up more money for debt repayment and savings.” 

Both ChatGPT and Mondoux suggested that I continue to prioritize retirement savings and increase contributions whenever possible to keep the momentum going. 

ChatGPT ended with a disclaimer about how realistic my homeownership goal is according to the facts I provided. 

“Based on your current savings account balance and recurring monthly expenses, it may be challenging to save enough for a $500,000 home purchase within the next 5 years. However, by prioritizing debt repayment and reducing expenses, you can make progress towards your goal.” 

How ChatGPT stacked up against our expert 

While ChatGPT generated a lot of the same advice as Mondoux, the process of working with a CFP to craft a financial plan toward homeownership provided greater value, which experts say isn’t a shock. When working with CFP, I was able to ask specific questions about how my plan might change over time and how to balance competing goals.

“AI is sometimes biased and wrong. It is primarily a function of the data it was trained on, which is why fine-tuning may be especially necessary in domains such as financial advice,” says McIntyre. “Recent advances in AI involve taking such models and further subjecting them to ‘fine-tuning’ with human feedback, wherein humans teach the AI how to perform useful tasks and rank responses. This teaches the AI how to leverage its core language knowledge to perform more useful tasks.” 

I was surprised to learn that ChatGPT was capable of outlining similar steps as a CFP might to help me hit my homeownership goal. From tackling high-interest debt, to accounting for other long-term goals like saving for retirement, it was able to provide a straightforward outline of the steps I might take to hit my goal. 

However, it lacked in a few places: 

  • ChatGPT did not account for potential changes in my income. One of the key conversation points when working with Mondoux was about how to adjust my plan depending on how my salary will change over the course of the next five years. On average, most companies will offer employees a 3%-5% salary bump per year to account for inflation. In five years’ time, this could amount to a 15% difference in my salary from the time I began preparing to buy a home, which could change how much I’m saving each month and alter my debt repayment timeline. 
  • ChatGPT’s limited knowledge of events after 2021 didn’t account for changing interest rates. While the software did call out the fact that credit card debt tends to carry higher interest rates and recommended paying off that debt as soon as possible, it couldn’t account for variable APRs and savings APYs based on Fed rate increases or student loan debt that may be on pause or not accruing interest as a result of the ongoing student loan moratorium. Whereas a CFP could take a closer look at current interest rates and give you clear direction on which debts to prioritize. 
  • A real human can help you set realistic goals. When creating this financial plan with Mondoux, the very first point of conversation was around the goal itself and how realistic it was according to the facts I provided. “I always like to start with the end in mind,” says Mondoux. “At the end of this, we have a very real goal that this client has, which is the purchase of a roughly half a million dollar home in five years. So the first thing I want to try to do is see how we can work that in and given the rest of the data is that goal realistic.” 
  • ChatGPT assumes that you know exactly how much you spend. ChatGPT only has the information you provide to go off of and use as the basis for your financial plan. You can ask the software follow-up questions, but it won’t know to ask you probing questions about your finances, which could lead to major gaps in information. “A lot of times people underestimate their expenses,” says Mondoux. “In this case, the person thinks that they spend $3,350 a month, but then we talk about ‘well do you go on vacation? Oh yeah, let’s take a $5,000 vacation a year,’ most people have these non-recurring expenses that aren’t appreciated in that month. Those are the other things that we need to make sure are accounted for, because if we find out that this person spent ten or fifteen grand a year on something that wasn’t in the budget that they looked at, we could be missing a big piece of the puzzle that AI software may not know to ask.” 
  • Sometimes, you need a real human to keep you accountable and adjust your plan if needed. Depending on your goal and how long your timeline is, your financial situation could change drastically and adjustments to your original plan may be in order. “I’m initiating that conversation, at least annually, potentially semi-annually to check back in as something materially changes, like a salary bump, a promotion, or an engagement. Those kinds of things always need to be factored in,” says Mondoux. “We’re heading to a destination, I might plug the address into my car GPS and it gives us a route but there’s always reasons to reroute and refresh the GPS. We always know the end location, but we’re always looking to reroute the plan if needed.” 

What experts have to say about using ChatGPT for money advice  

While monumental advances have been made in the AI space in the past few years, experts say it’s wise to take AI-generated advice with a grain of salt while the software is still being fine-tuned. On the one hand, AI could give consumers easier and more affordable access to financial advice. 

But as for the quality of that advice—it’s still up for debate. 

“It’s important to remember that ChatGPT was trained to be creative and to generate human-like text,” says McIntyre. “It wasn’t necessarily trained to be right, as preeminent machine learning researcher Andrew Ng amusingly pointed out.” 


This website uses cookies. By continuing to use this site, you accept our use of cookies.