As the age of Artificial Intelligence (AI) advances further and further, it seemingly has no limitations. Or at least no cap on its future potential.
The rising popularity and use of Large Language Models (LLMs), such as ChatGPT, Claude, and Gemini, have prompted (no pun intended) individuals to increasingly rely more on these bots for everyday tasks.
If you’ve been living under a rock and are unfamiliar with LLMs, they operate in a manner similar to a personal assistant. A personal assistant with unlimited knowledge, and someone who never disagrees with you. You can input prompts, which help guide the program to spew out the specific result or information that you are after. The possibilities are endless. For instance, you can use an LLM to help build a marketing plan, edit a website that you’ve built, or gather information from the internet in seconds. And its capabilities are growing.
It would be naïve to believe individuals wouldn’t eventually attempt to use LLMs for investment advice. Especially for financial DIYers, it would ostensibly be a no-brainer. As we’ve mentioned before, “beating the market” has been a long-standing puzzle that investors have tried to solve.
But the larger question becomes, “What is the efficacy of these LLMs related to stock investment advice?”
Well, wonder no more!
A recent study conducted by the National Bureau of Economic Research sheds some light. The study used different popular LLM models on the market and tested the results of seeking investment advice over a period of time. The researchers noted that “this study focuses on how AI platforms might service an untrained household investor, not a professional who might refine queries in more precise ways.”
They used two different types of stock portfolios – the first was passively managed (buy and hold), and the second was actively managed (could change its stock picks daily).
For context, below are the LLMs that they used for this study:
· OpenAI’s
· ChatGPT 5.0 and ChatGPT 5.2
· Anthropic’s Claude Sonnet 4.5
· Google’s Gemini 2.5 Flash
· xAI’s Grok 4.1 Fast
So how did the LLMs do?
While some of the data demonstrated some positive outcomes of using the LLMs for investment advice, the researchers were less bullish on the results.
The study found that the recommendations provided by the LLMs were mostly based on the media coverage received. Selected stocks had almost 10 times more news articles than the average Compustat firm. The researchers noted, “… the ability to grab attention within the universe of corporate news is a major driver of AI’s recommendations.”
It was noted in the study that the portfolios outperformed the S&P 500 over the length of the study, but did not earn abnormal returns when considering trade costs and the favoritism towards AI/Tech stocks. “AI takes risk, recommends a narrow set of assets, focuses on specific industries, and does not appear to exhibit better performance than passive characteristic-based benchmarks.”
The study also observed that the LLM portfolios became more concentrated over time. “… as ChatGPT 5.0 actively manages the portfolio, it becomes even more concentrated. This pattern also appears to be true for Claude Sonnet 4.5, Grok 4.1 Fast, and Gemini 2.5 Flash. So, AI primarily recommends holding large, successful tech stocks, and many sectors of the economy are not represented in the portfolios at all.”
Researchers concluded that, “Based on the preliminary results in this paper, it appears that more oversight is needed to assure that people do not misuse this powerful source of information and experience welfare losses.”
Overconcentration in AI stocks
AI has exploded, and stocks related to AI (e.g., chipmakers) have seen enormous growth over the past few years. So, it’s not overly shocking that the LLMs picked more of the AI stocks for the portfolios.
As of today, 14 companies have at least a $1 trillion market cap. Micron is the latest after achieving this mark last week. Out of the 14 companies, 12 are heavily involved in producing AI services, many of which are an integral part of the AI infrastructure (e.g., chip manufacturers).
Yet, being so heavily concentrated in one sector comes with potential risks, known as concentration risk. Assets within the same sector tend to be highly correlated – if the AI sector experiences a major downturn, the stocks/companies within it will likely fall together. Having a more diversified portfolio helps hedge against the overconcentration risk.
There is also growing concern that we are currently in an AI “bubble” and that bubble may burst at any time. It draws some similarities with the dot-com bubble of 2000.
Financial DIYers and delegators
In the financial service industry, we typically separate consumers into two categories. The DIYers and the delegators.
Financial DIYers will always look for ways to seek less expensive advice and have confidence that they can manage their own finances. They may utilize robo-advisors rather than a traditional advisor.
A delegator typically wants to pass off their finances to someone else. This may be due to a lack of knowledge, no free time, or they want someone to help them make decisions.
Oftentimes, financial or wealth advisors are just seen as handling the investments. The benefit of working with a financial advisor is not necessarily the stocks they pick, but the ancillary benefits. Studies conducted by Vanguard, Morningstar, and Evestnet point to significant value (upwards of 3% or more per year) that a financial advisor can provide, not just related to investments.
This is achieved through tax planning, thoughtful portfolios constructed based on the client’s needs and goals, and, maybe more importantly, accountability to stick with a plan and behavioral coaching. One’s relationship with money has a huge impact on their financial outlook and long-term success.
So, financial planning and advice are much more than investments. In fact, oftentimes the investment strategy is set in place, and more pertinent items come to the forefront. Also, unplanned events happen in people’s lives frequently. You may lose a job, have a family member get sick, or face any number of unforeseen circumstances. This is where a financial advisor provides the most value. To help you navigate all of life’s ups and downs.
There are also fears around the effect of using LLMs as your “human interaction”. A research article in the Proceedings of the National Academy of Sciences (PNAS) cautioned against using LLMs as a “human surrogate”. In other words, replacing your interaction with humans with interactions with LLMs.
In closing
While I think there’s tremendous value in working with a financial advisor (I’ve experienced it personally), I’m not an AI hater. I’ve used LLMs, and although they have their limitations, overall, they have been useful.
Do I think that AI will be able to advise more appropriately regarding portfolio construction? Yes.
Will AI replace our need and craving for human interaction? For some, perhaps. But for most, LLMs will have a difficult time replicating the complexity of being human. Our emotions, thoughts, and feelings cannot be duplicated overnight — and may never be.
*This is not meant to be investment advice and is for general information purposes only.
Did you know?
55% of Americans used AI to aid their financial management decisions in 2025 — up from just 10% in 2024. However, only 18% of Americans say they would trust AI to make financial recommendations on its own — underscoring that human involvement remains central.
Something to ponder…
Would you put full trust in an LLM to provide comprehensive investment and/or planning advice? Why or why not?
In case you missed it…
- The inaugural Enhanced Games took place on May 24. Think of the Olympics on steroids, literally. Athletes were permitted to use performance-enhancing drugs (PEDs) and offered a $10 million prize for anyone who could best Usain Bolt’s 100-meter world record. Fred Kerley won the 100-meter dash in 9.97 seconds, well short of the 9.58 mark set by Bolt in 2009.
- The Ferrari Luce was revealed this past week as the car company’s first electric vehicle. Scheduled for release in 2027, it boasts 1,035 horsepower and 4 battery-powered motors. Not everyone was excited about the sports car manufacturer’s foray into the EV market. Ferrari’s stock dipped 8% the day after the release, and former Chairman Luca Cordero di Montezemolo stated the model risks “destroying a myth.”
- Paris Saint-Germain (Pah-ree San Zher-mahn) won the UEFA Champions League title on Saturday, defeating English club Arsenal 4-3 on penalty kicks. They became the first club to go back-to-back since Real Madrid did so in 2016 and 2017.
- Team Canada fell to Norway 3-2 in overtime, missing the podium at the hockey World Championships. This result comes on the heels of a heartbreaking loss to the U.S. in the Milan-Cortina Olympic gold medal game. It’s been a rough 2026 for Canada. Not only are they showing inferiority in their national sport, but they also fell into a “technical” recession following a contraction of real gross domestic product after quarter 1 of this year. For those wondering, I’m Canadian.