AI Point/Counterpoint
The debate over artificial intelligence (AI) is just getting started, but there is considerable buzz about its potential impact on the financial services industry. In this article, Matt Cohen and Kristan Anderson debate questions related to AI and find some common ground on a few of the more pertinent issues that will impact clients and how we work with them.
1. Will using AI in financial planning have an impact on client privacy?
MC – I like to think not, at least in relation to the current privacy risks out there today. If client financial planning data is being used in large language models to “teach” AI systems, I have to believe that there exists a cyber security ecosystem around those AI tools to make the data anonymous, prevent data breaches, etc. The benefits of AI need to outweigh the risks though.
KA – As with many planning-related questions, I think the answer is, “that depends.” Matt makes a good point about using client data to inform or teach the model. But if the client isn’t aware that their data is being used and practitioners don’t do their due diligence regarding AI platforms, security, access to data, and don’t take steps to keep client data anonymous, then it can be a real concern. If I learned nothing from my classwork in information security, it is that humans are usually the biggest security threat on the internet.
2. Do AI systems need to be explainable to clients?
MC – When brand new technology is used, the way it is being used needs to be explained up front. Right now, AI can transcribe a call and produce notes from it, along with suggesting some action items. We don’t use this technology today, but it’s something we might use in the future. I would absolutely disclose to clients the transcribing tool up front and provide an opt out if they are uncomfortable. In 5 years, who knows, maybe it’ll be assumed and socially acceptable that every video chat we do is being transcribed.
KA – I have a different take on this in that if I’m using a tool to help me do the work for a client, then I need to understand how it is helping me so that I can explain the outcome to a client as needed. The so-called “black box” solution to a problem is convenient and efficient but rarely can be applied to even a simple client question without caveats or the need for additional context. Understanding that you can always pose follow-up questions or qualifications, at what point are you just directing traffic, versus doing the work in the first place? And if you are doing the work yourself, I would argue that you are in a better position to explain and defend the thought process. I agree with Matt that transcription tools will be a positive addition to the process of gathering data and documenting meetings. I don’t need to know how it works, just that the notes are getting done.
3. Are there biases in AI tools or solutions?
MC – Humans have biases, and AI is learning from humans, thus there must be biases in AI. I recently asked ChatGPT to build a portfolio for a 65-year-old, and it suggested a moderate risk profile with 50% invested in equities. That’s the result of decades of groupthink in our industry that a 65-year-old must be a moderate investor. But what if this investor had enough pension income to not need any cash flow, and wish to build a legacy for their kids? What if their spending rate was sufficiently high where they needed a more aggressive portfolio to maximize the return potential? Those are all the follow-on questions we would ask, but off the bat ChatGPT was biased towards a moderate portfolio without knowing anything else.
KA – I agree that there probably isn’t a way to remove bias from planning, whether it be from humans or AI tools. However, I feel like an AI tool is going to have bias that you can’t get around by simply throwing more parameters at it. At some point, the human element, as flawed as it may be, will need to use its ability to think critically without emotion to provide solutions.
4. Is AI going to transform the financial services industry?
MC – AI will impact financial services, just like most industries, it’s simply a matter of how. Much of the advice we provide stems from a series of iterative steps, with intake questions, follow up refining questions, etc. Machines are very good at this. Take the portfolio example in the last question. It doesn’t seem like a stretch to think that a software vendor could populate a series of questions to recommend a more tailored portfolio. We see this right now with vendors like TurboTax, e.g. “Did you make a nondeductible IRA contribution?” and if “yes” it knows to include form 8606 in the tax return. How hard would it be for TurboTax to review some key data points to prompt a few refining questions to then answer, “Should I make the nondeductible IRA contribution?”
KA – I’ve been working in financial planning for 30 years now and I’ve seen new tools come into the picture and been skeptical of their impact. Back in the day, Monte Carlo analysis was the next big thing. The idea was that relying on straight line growth of investments to assess whether spending would outpace investment growth was not rational and you needed to include the impact of market volatility. And while Monte Carlo does that, it also may cause the advisor to artificially rely on hitting a certain target number, versus really knowing what is going on in the cash flow over time. I see AI in the same way. I’m not sure what the applications are going to be and I look forward to working these tools into the process, but at the end of the day, there’s no substitute for getting into the weeds with a client’s data and the reported outcomes.
5. Can AI tools be better financial advisors than humans?
MC – I am going to give a biased answer here and say probably not. Much of the advice we give is choosing the best option from a few good options. However, what makes the “best option” the most optimal is often subjective. Sometimes the “math” will point to a certain outcome, e.g., doing a Roth conversion in a low tax year. But what are the practical implications for cash flow and investments as well as client preferences for paying taxes? There really isn’t a “wrong” choice here, but will ChatGPT reassure this client that there is nothing problematic with their choice to forgo the Roth conversion?
KA – I think we agree here, although I feel like ChatGPT will be somewhat obsequious when it comes to assuring people that they are making good choices, regardless. I’m a firm believer that numbers can tell whatever story you want them to tell. So, it does require an element of knowing that the story you are telling is consistent with the client’s needs and wants and the information that you are given. Time will tell if AI learns from the best of us and becomes a partner in the process.
6. Will AI be geared towards the practitioner with better tools to serve clients, or will AI replace some services our industry provides?
MC – The tools AI can provide practitioners are what excites me the most. As I mentioned above, I think the benefits of AI will be tools that let us better serve clients rather than services we provide being automated and self-directed. For example, we may be able to write descriptions about our clients in CRM, along with meeting notes, and AI could scrub all of that and make recommendations for us to review. It might catch the fact that a client doesn’t have umbrella insurance but recently had a large increase to their net worth. While we might have made the same recommendation as part of our regular meeting review, it’s not a stretch to see how AI could catch this sort of thing in real time.
KA – As a skeptic, I’d say that AI tools for practitioners to use will be the most likely outcome, rather than a full replacement of the advisor. I’ve seen how robo-advisors have been adopted and think that most people want to develop a relationship with their advisor, if only to have a person to call when things are not going as expected.
West Financial Services, Inc. (“WFS”) offers investment advisory services and is registered with the U.S. Securities and Exchange Commission (“SEC”). SEC registration does not constitute an endorsement of the firm by the SEC nor does it indicate that the firm has attained a particular level of skill or ability. You should carefully read and review all information provided by WFS, including Form ADV Part 1A, Part 2A brochure and all supplements, and Form CRS.
This information is intended to be educational in nature, and not as a recommendation of any particular strategy, approach, product, security, or concept. These materials are not intended as any form of substitute for individualized investment advice. The discussion is general in nature, and therefore not intended to recommend or endorse any asset class, security, or technical aspect of any security for the purpose of allowing a reader to use the approach on their own. You should not treat these materials as advice in relation to legal, taxation, or investment matters. Before participating in any investment program or making any investment, clients as well as all other readers are encouraged to consult with their own professional advisers, including investment advisers and tax advisers.