As generative AI and machine learning kick into higher gear, industries across the board are scrambling to adopt and integrate ChatGPT and other large language models (LLMs) into their workflow to enhance speed and efficiency.
One such domain is the financial services industry. Perhaps one of the most conservative of all sectors, it has now started to warm up to the potential of generative AI. This could have meaningful implications for investors and the broader financial planning industry.
Behemoths on the bandwagon
There is no better testament to generative AI’s growing impact than global investment banks such as JP Morgan, Morgan Stanley and Goldman Sachs recognising the potential of AI and embracing the technology wholeheartedly.
These institutions are leveraging new AI models to develop sophisticated tools that help financial advisers respond to client queries and create personalised investment decisions.
While the fast-shrinking pool of sceptics often points to inaccuracies, experts argue these powerful AI tools would continue to become more sophisticated and are poised to revolutionise the financial planning landscape.
ChatGPT comes out on top
Although the role of generative AI in the financial industry is in its infancy, there are experiments and studies that provide evidence of the technology's true potential.
For instance, JP Morgan is developing a ChatGPT-like AI service, called IndexGPT, a trademark filing shows, that can select securities and investment advice.
According to the filing, IndexGPT plans to utilise cloud computing software powered by AI that can carry out a “selection of financial securities and financial assets” and select investments “tailored to customer needs”.
Not to be outdone, Morgan Stanley is currently conducting trials of a chatbot powered by OpenAI technology for its 16,000 financial advisers.
Like ChatGPT, this tool will provide instant answers to advisers' questions. However, it will generate responses exclusively from the approximately 100,000 vetted research pieces from Morgan Stanley, reducing the likelihood of errors.
AI may be some way away from handling money independently, but a study conducted by two South Korean academics shows that the technology beat out random selection in creating a diversified portfolio.
As an LLM, ChatGPT has the potential to capture special aspects of the market that human investors cannot, rendering it a valuable tool for managing portfolios, the study says.
It notes that ChatGPT can identify abstract relationships between assets, particularly in terms of their dissimilarity regarding asset classes, which helps to achieve better diversification than an average portfolio.
In another test, a simulated portfolio of stocks chosen by ChatGPT outperformed several top-performing investment funds in the UK.
As of August 25, the ChatGPT fund had grown 9.63 per cent, comprehensively outpacing leading UK funds that collectively managed a meagre 0.43 per cent gain for the same period.
While the fund continues to outperform, 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 tweaks the portfolio on an ongoing basis,” says Jon Ostler, chief executive of Finder.com, the company behind the study.
New AI tools are much more efficient at processing and analysing large amounts of data to detect patterns.
“Generative AI has the potential to support and enhance many aspects of financial planning if it has access” to the most recent and relevant financial data, says Mr Ostler.
Notably, AI is better than Monte Carlo simulations; while the latter uses models constructed by experts to predict probabilities, the former builds its own models to predict future outcomes, Mr Ostler adds.
When it comes to stock markets, “the lack of bias in AI could give it the edge, [and thus] could be a real game changer in the future”, he says.
Michael Zagari, a Montreal, Canada-based associate portfolio manager with Mandeville Private Client and Zagari + Simpson, is one of the early adopters of ChatGPT.
“The tool helps me point to where I should invest more time researching on a particular data point of moment in time,” says Mr Zagari who uses ChatGPT to understand investment trends.
“The trade off is the data I am receiving has a time lag,” he adds, referring to a key drawback of ChatGPT whose data set is limited up to 2021 and has no awareness of events that have occurred since then.
Watch: The AI revolution: What does our future look like?
AI and financial planning
More recently, AI models have become highly sophisticated at making predictions and working with different scenarios to test financial outcomes.
While they can’t foretell the future – yet – predictive AI tools can play out many different scenarios for investors, providing the best possible outcomes for them.
“There could be huge cost and efficiency benefits from using AI as it can sift through vast amounts of data both quickly and accurately,” says Mr Ostler. “After the hype around ChatGPT calms down a bit, simple automation tasks like this may end up being the biggest impact of AI.”
ChatGTP is a valuable asset that supports advisers, not supplant them, says Mr Zagari.
“I’m not convinced that people will fully trust the advice they receive from AI, which is no different than people trusting their self-driving cars with an autopilot drive,” he says.
That could change in the coming years. The constantly improving quality of AI’s predictions have prompted some to believe that, in the not so distant future, the technology could make better decisions than humans.
“Even the best finance professionals have inherent biases that will impact their judgment and AI may have an opportunity to remove this from their processes,” says Mr Ostler.
Still, AI will not make anyone rich and likely will mirror market/index returns, which is great from a cost perspective, Mr Zagari says.
“If your retirement plan requires you to take on more risk because the investor is not prepared to save more after tax, a shortfall in financial goals will remain in play.”
Some financial industry experts prefer to see generative AI more as a co-pilot that can assist in some areas of a financial adviser's job.
GPT-4, a more advanced version of ChatGPT, can parse vast amounts of data, analyse multiple variables and offer possible scenarios.
Although its prowess doesn’t extend to crystal-ball predictions, “it can aid in creating hyper-personalised strategies” drawn from specific input, says, Dave Mazza, chief strategy officer at Roundhill Investments, a company that has begun to integrate generative AI into many of its workflows to take advantage of its key benefits, such as accuracy, speed, and cost-effectiveness.
In time, generative AI could become increasingly advanced and adapt at breaking down more complex areas of financial planning and “offer more personalised advice”, Mr Mazza says.
As with all emerging technologies, generative AI has its limitations and requires additional refinement. Potential risks include data privacy and security, the ethical application of AI and algorithmic biases, among others.
“If AI is managing hundreds of thousands of dollars' worth of investments, a small mistake or misunderstanding could wreak havoc on portfolios,” cautions Mr Ostler.
Some of these risks could be mitigated with proper design and supervision, says Mr Mazza, who admits that “the role of advisers may shift to interpreting generative AI outputs and providing more personalised advice”.
Another issue is accountability. Who is ultimately responsible for the advice given?
“I assume that will remain the financial adviser’s responsibility, just as it’s the adviser’s responsibility today for overseeing the computer-generated information that he or she provides,” says John Rekenthaler, director of research for Morningstar Research Services.
AI can give financial advisers an edge
As things stand, ChatGPT and other LLMs in their current iteration remain an auxiliary tool. “ChatGTP acts as a complement to my research and not as a replacement, which is similar to how investors should view the technology,” says Mr Zagari.
In its role as a personal assistant, AI could perform many mundane tasks for advisers, allowing them to focus on other areas of financial planning and client engagement.
As the technology becomes advanced enough to grasp more intricate aspects of financial planning, Mr Rekenthaler is convinced that it will provide an edge to financial advisers who are able to effectively integrate AI technology into their workflow.
“Across many industries, generative AI has the potential to make even novices behave like power-users,” Mr Mazza notes, stressing that “in the near future, advisers who embrace AI may have a leg-up on those who don’t”.