One of the problems with building a nest egg is that we generally don’t focus on doing this early enough and build up the practice of saving before getting into our 40s. By then, it is a bit too late. But when we first start out, we don’t have enough money to invest unless we receive it as a gift or inheritance to justify working with a financial planner or stockbroker to set up a long-term personalized engagement.
This creates a cart and horse problem. If we don’t start saving early, we won’t get the advice we need until after we’ve saved enough. As a result, many of us don’t save enough. This is one of the benefits of the new generative AI platforms based on large language models. They have the potential to engage deeply with potential investors, but the cost of that engagement is low so they can be used early in a person’s career profitably and help young investors get into the habit of investing before it is too late.
This creates both an opportunity and several problems. Let’s discuss both this month.
Opportunity
The opportunity for AI use in investing is that the cost of the advice can be minimized to make it financially viable for those with minimal investment dollars. In addition, it can engage deeply and keep the clients informed, provide real time warnings and advice at any time during the day or night, never get frustrated, and can know more about every individual investor than a human advisor can remember.
So, the best opportunity is to use this technology as an on-ramp to a firm’s investment and brokerage services by developing both a relationship with this new young customer and their related habits that will grow the investment over time into something that would warrant human support.
In addition, while the training set to create such an AI is expensive (generally estimated in the $100M range), it could then be sold to many small investment houses, which would spread the cost across many customers and make it far more affordable. This would allow smaller firms to compete with the larger firms that can afford to create this training set themselves.
Current development of generative AI in the financial industry isn’t just in personalization or the customizing of financial advisory services for individual clients at scale. The technology is also focused on automating financial reporting, risk assessment and cyber security. An example of a generative AI application is Fin Genius which can take in information on the client’s income, financial goals, risk tolerance and current assets. It then takes that information, factors in market growth percentages, financial conditions like interest rates and taxes, inflation, and where the current investment trends are to create a customized investment report that will optimize an investment strategy based on all those metrics. This report will also consider the investor’s spouse and kids to craft a result that will provide for the future needs of both.
Problems
Like any technology, this technology can be abused and biased with similar outcomes, but better evidence trails for those seeking to abuse this technology to take advantage of customers at scale. Individuals, however, need to focus on assuring that the service, human or digital, is trustworthy, knowledgeable (negligence can be as or costlier than a crime), and has protections against being compromised (does the firm tend to cover up or report and correct problems?).
AIs operate at machine speeds. This their mistakes could result in damages far greater than any human could create (and humans have made some huge mistakes over the years, the sub-prime mortgage collapse being a case in point).
So financial institutions that have a history of making bad mistakes should likely be avoided like the plague if you are considering using one of these new AI tools. Finally, AIs, like any tool, aren’t inherently good or bad. That depends on the integrity and quality of their training set, and how the tool is used. This last not only again suggests you want to see a history of ethical behavior but that the firm providing the service understands it. This could be problematic for a small investment house without the ability to take a course like Udacity’s “Machine Learning Engineer” course. (The cost is around $750 per student).
Wrapping up:
One of the markets that has embraced AI heavily is the financial market but the cost of spinning up a generative AI is prohibitive, meaning that it will initially only be available to large firms and eventually as a cloud service that could go through a smaller firm or bypass them. (I can imagine that after going into the healthcare business, Amazon Web Services may next move into the financial market). Competing with these larger AI firms will require that smaller firms either collectively spin up or use services that can level the playing field or face declining potential markets as companies like Amazon do to them what they have done to bookstores and brick and mortar retailers.
To prevent migrations, it would be wise to make sure your customer relationships are solid and that you work to spin up competing services to hold customers, arguing that you can deal with them more effectively than some faceless company. However, this will only work for a while because these AI solutions are advancing at unbelievable speeds. Those firms that survive will the ones that best understand this new technology and can use it effectively to protect and grow their customer base.
