ChatGPT software and other generative artificial intelligence tools are muscling their way into the financial services industry, and will be involved in both retirement planning and constructing investment portfolios.
JPMorgan is developing a ChatGPT-like A.I. service, called IndexGPT, according to a trademark filing, that can select securities and investment advice.
Rival Morgan Stanley is also testing an OpenAI-powered chatbot for its 16,000 financial advisors to help better serve clients. Like ChatGPT, this tool will provide instant answers to advisors’ questions, drawing on Morgan Stanley research.
ChatGPT Makes Better Investment Decisions
The technology isn’t yet running money on its own, but a study conducted by two academics in South Korea shows a portfolio constructed using ChatGPT outperformed random stock selection as a portfolio manager. Among other things, ChatGPT was better at picking diversified assets, producing a more efficient portfolio.
In another experiment, a dummy portfolio of stocks selected by ChatGPT significantly outperformed some of the leading investment funds in the U.K. As of March 6 to April 28, the continuing study showed that the AI-generated portfolio increased in value by 4.9%, surpassing the 3% gains of the S&P 500 index, while major U.K. investment funds lost 0.8%, over the same period. As of July 27, the ChatGPT fund had racked up nearly a10% return.
AI considered key principles from top-performing funds to select personalized stocks. “This process would be very difficult for an amateur investor, and could easily be derailed by conscious and unconscious bias,” says Jon Ostler, CEO at Finder.com, a global fintech firm that conducted the study.
While the fund continues to outperform, Ostler admits it doesn’t yet have access to real time information. The next step would be to make a portfolio that constantly monitors the market and continually tweaks the portfolio based on external factors, he adds.
“AI is fantastic for synthesizing large amounts of data,” says Ostler. “In theory, generative AI has the potential to support and enhance many aspects of retirement planning if it has access to up-to-date and specific financial data sources and analyst research.”
AI models are getting good at predictions and simulations which can be useful in testing different future scenarios and their impact on specific financial goals. “AI could be used to develop, test and illustrate retirement plans quickly, as long as all the individual circumstances of a person can be fed into the model effectively,” says Ostler. Unlike Monte Carlo simulations that use models constructed by experts to predict probabilities, AI builds its own models to predict future outcomes, Ostler says.
Generative AI also holds the potential for making the retirement planning process more efficient. “The use of AI-powered automation will allow retirement plans to be continuously adapted based on changing circumstances and new data,” Ostler says. “A plan could, thus, be designed by an advisor using AI and then updated and enhanced automatically with little human effort.”
Models like GPT-4—a more advanced version of ChatGPT—can analyze vast amounts of data, consider multiple variables, and generate possible scenarios. While it can’t predict the future, “it can aid in creating hyper-personalized strategies based on the user’s input and the data it has been trained on for these purposes,” says Dave Mazza, chief strategy officer at Roundhill Investments.
For example, a client may have dueling objectives such as needing current income for living expense and capital appreciation for the future. AI could help serve as the advisor’s co-pilot in analyzing the range of acceptable outcomes and determine what is and isn’t relevant to a client’s individual requirements and craft personalized strategies with greater customization to better meet their investment objectives, Mazza notes.
His investment firm is in the early stages of incorporating generative AI into numerous workflows to gain additional precision, speed, and cost-effectiveness. “These AI models can process massive data sets in seconds and provide personalized advice,” which could augment advisors’ productivity and optimize their business, he adds.
Over time, generative AI could acquire a fine-grained ability to understand more complex aspects of retirement planning, such as dynamic portfolio management. Generative AI might evolve to develop personalized investment strategies that flexibly respond in real time to changes in an individual’s financial circumstances and market dynamics. “There are expectations of advancements that would enable generative AI to understand user emotions, needs, and aspirations more accurately to offer more personalized advice,” says Mazza.
All told, ChatGPT is a complementary tool. As a personal assistant AI could perform many routine tasks of advisors, leaving them with the responsibility of reviewing the AI’s recommendations and providing the final stamp of approval.
“AI could permit financial advisors to be far more efficient,” says John Rekenthaler, director of research for Morningstar Research Services. In the future, advisors may be less valued for their deep knowledge on a subject, as AI programs can replace that knowledge. Instead, they may be more valued for their ability to effectively use AI technology in their work, he adds.
Rekenthaler says AI’s role will grow. “Further down the line, AI will become intertwined with the financial planning process,” he says. “The advisor will retain the personal relationship, but AI will assist in asking the questions and will ultimately create the financial plans.”
Write to editors@barrons.com
Read the full article here
