Can AI Help Manage Your Portfolio?

About 40% of consumers “have sought help from [artificial intelligence (AI)] to manage their personal finances,” according to a recent JD Power survey. More than one in four respondents (27%) found AI to be “somewhat or significantly helpful.”

The top three ways consumers have used AI are to compare prices (24%), to find coupons or discounts (22%) and to get suggestions for increasing income or saving money (21%). I can vouch for using AI to compare prices. I recently used ChatGPT to help me get a better price on a Garmin running watch. The result was $25 in savings.

The JD Power survey follows a Lloyds Banking Group survey that found that 56% of adults in the U.K. used AI in the last 12 months to assist them with managing money. More than half of those individuals sought information on “budgeting, savings planning or general financial education.” Additionally, over one-third of those AI users (37%) reported that they “engage with AI for investment research and recommendations.”

Is AI reliable enough to use for managing your portfolio? The answer is “it depends.” What information you ask it for and how you write queries (aka prompts) matters greatly. So does your ability to confirm whether the answer it provides is correct.

Demographics and Prompts Influence AI’s Suggestions

A study conducted earlier this year by researchers at the Massachusetts Institute of Technology (MIT) and Stanford University asked a demographically representative sample of adults to write free-text prompts to a large language model (LLM) financial adviser. Each person was asked to describe their finances, then ask the AI chatbot how much to spend and how to invest between stocks and safer options.

These prompts were then organized into 12 buckets defined by employment status, age and income. Prompts from each bucket were randomly selected and used to create simulated life paths. The researchers simulated 1,000 individuals. Each person started at age 22 with no wealth, retired at age 65 and died at age 90. ChatGPT version GPT-5.2 was used.

The chatbot recommended saving more during one’s working life and drawing down savings in retirement. This is consistent with well-established financial practices. The AI-generated allocations called for an average 65% weighting in stocks that decreases with age. The one exception was for young adults with little wealth; the AI chatbot suggested that they build an emergency fund first.

About 30% of suggested savings rates were multiples of 10%, and 34% of suggested dollar amounts were multiples of $5,000, demonstrating a bias toward simple, round numbers. The 4% rule was proposed as the basis for withdrawal strategies in 85% of the chatbot’s recommendations. This rule calls for withdrawing 4% of your portfolio balance at retirement, then increasing that dollar amount each year for inflation. The study’s authors noted that academic prompts led to more flexibility in determining how much to withdraw.

The prompts used mattered. Modeled wealth at age 60 was lower for women, study participants who had low levels of financial literacy and those who had not previously used AI for financial advice. According to the study’s authors, the AI chatbot suggested lower stock allocations for female investors and those with lower financial literacy. The chatbot’s simulations also resulted in lower savings for participants without the relevant AI experience. All outcomes assume that investors will follow the AI’s advice and act accordingly—a big assumption.

Where AI Can Help Investors

AI chatbots are good at summarizing long documents into bullet points. This can be useful if you want to analyze a company’s regulatory filings, investor presentations or other lengthy documents.

I’ve found AI to work well with spreadsheets. It can quickly convert a formatted PDF table into an Excel file. You can use it to create spreadsheet models to do your own analysis or forecasting. However, you will have to go through the spreadsheet to ensure that all calculations are done properly, especially if data is being pulled in from different cells. The more information and feedback you give the chatbot, the better the outcome.

You can also use AI to search for information, whether on a specific investment, economic data or financial research. In doing so, be sure to instruct the chatbot to only look at reliable sources and then do your own research to confirm that its output is correct.

As far as your allocation is concerned, AI can suggest a portfolio mix, but view its suggestions as a starting point instead of personalized advice. The allocation that ChatGPT gave me this morning is more conservative than what I use for my retirement savings. AI chatbots have shown themselves to be reward-seeking and therefore can be prone to giving you answers they think you want to hear instead of answers you don’t want to hear.