Case Study in AI Finance – Ep. 292

Welcome to the Accumulating Wealth podcast. I’m Hunter Satterfield. And I’m Judson Crawford. We’re CPAs, wealth advisors, and partners at Cain Watters and Associates, a financial services firm here to help people navigate the decisions they face every day. You know, Judson, a new survey reports that 66% of Americans, not 67, 66% of Americans who have used generative AI are using it for financial advice, and the percentage gets even higher for Gen Z and millennials. 

It’s fast, accessible, and it can be helpful, but like medical advice, we have to know when to use it and when to consult a professional. That’s right. Us old guys are not professionals, so today we’ve invited three of our Gen Z advisors to join us on the pod to share their experience testing out what it would be like if AI was their financial guide. 

Yeah, I’m excited for this. Let’s go. 

Okay, so Hunter, today we’re going to talk a little bit about AI, something we’ve never, ever even touched on the podcast. And then we’re going to do PPV and interest rates and ERTC. That’s right. Okay, so, why is it that we even need to talk about AI on this financial podcast? Well, because it’s kind of a thing. In fact, by the time we drop this recording, things will have changed already in just four days. 

100%. And it’s interesting. We’re actually getting clients that are coming in. We talked about this with FinTok, clients that come in with stuff from TikTok or whatever, and they’re like, “Hey, Judson and Hunter, or whoever, I’ve learned this out there. Tell me about it.”  

And now we’re starting to get clients that are like, “Hey, I read about this on Chat- ChatGPT-” 100%. “… different tax advice or whatever else,” and we’re seeing- some of it’s good. I was going to say, maybe it’s good, it’s not just TikTok anymore. Yeah. Good point. 

But some of it is good, but then there’s some flaws to it as well. And we have now seen, listeners, multiple clients that have come in and said, “Hey, I found this on some chatbot, and I’m now going to do this,” and it might be a moderately aggressive… not even moderately aggressive, just a moderately advanced tax strategy, and the chatbot’s just wrong. 

Yeah. And it takes us walking through with the client on why it’s wrong. But yeah, so we thought we’d be good to cover this. Well, and you mentioned in our completely unscripted intro that there is a survey out there that says that 66% of Americans are using generative AI for financial advice. 

Yeah, and that was all the way back from September of last year. It was a Credit Karma Intuit Credit Karma survey data. So I mean, that number’s even higher now, right? And we’re seeing clients that are maybe new clients, prospective clients that call in and say, “Hey, why wouldn’t I just use a chatbot or a generative AI tool to be my financial advisor?” 

And it shows that folks that are out there are actually looking at it. There was a relatively new survey that just dropped by eToro that said 30% of everyday investors are using AI for their portfolios, right? So I mean, and I’ve seen this out there too when I’m on X or social media where it’s like, “Hey, Grok or ChatGPT, give me the eight names that I should be buying to take advantage of the oil crisis in Iran.” 

No fundamentals whatsoever, just, “Hey Grok, give me these,” and take it seriously. I’ve even seen that as a prompt. Take it seriously, Grok. Because that’s something we should do. So, long and short of it is that it’s a real topic. We could probably record a new one of these every few months and things would change, but we thought we’d have a little bit of fun today. 

So what do we do here? Okay, so here’s my question for you though, because I think this is interesting. Finance ranks as the second most used case for generative AI. You may have already read this. Did you already read this? I did not. Okay, good. Okay, so if finance is second, what type of advice is the number one that people are searching for? 

I hope it’s not dating advice. That would be great. No, it has to be is it the replacement for WebMD? Exactly. Yeah. Health and wellness. I have, “Hey, ChatGPT, I have this funny mole. I have a spot on my back.” Yeah. Is it really? It’s health and wellness? It is health and wellness. Yeah. Yeah. 

Next we can call some doctors on here and talk about how wrong it is. Well, have you used it for health and wellness? No, I have not. No workout- You’ve never put like a workout in or anything in there? Oh, I guess that would be covered. I thought you were talking about like “my heart hurts.” Yeah, I think so. 

Well, that would be health and wellness. I’m probably going to call. I think health and wellness would be. Yeah, yeah. You’re probably right. Yeah. You’re probably right about that. Most of my stuff is “hey Grok, who were the 10 best baseball players of all time?” Simple stuff like that. It’s all sports. It’s not high on the list, I guess. Yeah, I’m not showing you my history. 

Okay. What we’ve done is we mentioned in the intro is we’ve asked a few of our associate planners to go in and research a couple different topics, and we’re going to have them on, and we’re going to just ask them some questions about what they learned. And I think it will be really interesting and beneficial. 

So, Hunter, who are we starting with today? We are going to start with Caleb. Welcome, Caleb. Thank you guys. Thank you guys for having me. Yeah, it’s good. And again, we, we teed up Caleb on one specific topic, and then you want to tell folks… well, first off, tell them what topic we asked you to go look at, and then just your process before we actually go through what you found. Yeah. 

So you guys had asked for me to look at tax planning advice through AI. And so what I did is personally, I use Gemini on my own. So, I’m familiar with the process of how I enter things into Gemini. I’m familiar with the response that it gives me typically, and so that’s the one that I chose. 

So it creates a personal learning model tailored to each individual. And so I backed out of my own and went into an incognito browser and started using Gemini not logged in, so that way it would not give me answers that say, “Hey, it knows that you’re a CPA. It knows I work at Cain Watters.” It’s not giving me advice based on that prior knowledge. 

I went in there and asked it a couple of personal questions and stuff that our clients would find pretty useful. 

I am now fascinated to understand your personal learning model, Judson. I didn’t even know these things existed. Well, I just have conversations with my AI now. Yeah, it’s fantastic. All right. We’re friends. 

Okay, Caleb. So what questions did you ask, and then what results did you get? Again, folks, listeners, this is on taxes specifically. 

So first thing I started off with was, “Hey, I have a car. I saw on TikTok that I can write off my car against my personal taxes. How can I do that as a W-2 employee?” 

Okay. And what’d you get? It’s a good one. So I kind of went along the lines of, “I want to go mow a couple buddies’ yards, make a couple hundred bucks, and take that, say that my car is the business vehicle, and write that off on my taxes.” Started off, it was like, “Well, you have to be in the pursuit of profit for this,” which is what the IRS guideline says. 

I pushed back on it a little bit and I’m like, “Well, I kind of want to be more aggressive.” So then it starts to recommend me home office deductions on top of my car and creating this massive loss on my tax return. And not only did it recommend these things, after I asked it, “Hey, is this too aggressive?” It said that all of these deductions are highly encouraged, legislated, and legal. 

Wow. That’s like the little devil on the shoulder, isn’t it? Yeah. I love it. It’s just like, what does Caleb want here? That was my personal learning model. Yeah. Meanwhile, my personal learning model is like, “No, no, no, you can’t do that.” It has an alarm going off in the background. Yeah. 

No, that’s fascinating because again, that little pushback, and then you get to home office deduction, which Judson, we talked about on here, I mean, this would be like episode 10 at this point, like 200 episodes ago. Please don’t. Is one of the most audited of all of them, right? 

And it’s very, very difficult to justify, but apparently Caleb’s personal learning model was okay with it. Well, if you’re mowing lawns, I think that it’s very appropriate. Yeah. Home office would be just like the yard, right? Yeah. I don’t really know how that works. Okay, then where’d you go from there?  

Yeah. I was waiting for it to ask me about my risk tolerance at any point in this, and it never did. So I guess it just assumes I’m pretty risky. But after that, I went over to wash sales, which is you sell a stock 30 days after you buy it, and the IRS does not allow you to take that loss if you buy it within that 30-day period. 

So I was asking, “How do I get around this? I really like this stock. I want to sell it, take a loss, but I want to be in it for that rocket up over the next couple of days.” 

I have to stop you for a second. You’re a sneaky little guy, man. Just trying to, like, get it, “Hey, I really want to get out of this wash sale rule thing,” which we talked about on here because the tax loss harvesting is a serious deal. Yep. All right. So then you said, “I want to get around it. How do I get out of it? Because I love this high-flying stock.” What did it tell you? 

It actually recommended that I buy out-of-the-money call options while waiting on the wash sale to expire. 

So that’s what he did. Oh my gosh. Immediately after. I am currently about to hit the floor on… this is insane. Stupid Gemini just pushing Caleb where he doesn’t need to go here. Yeah. 

And not once did it ever explain the risks associated with out-of-the-money call options, and actually the IRS treats it as a like security, so it does not avoid wash sale rules. So A. I could have not avoided wash sale rules and lost my entire investment. 

And you would have done this from your home office- Yes … that you’ve deducted. Yeah, indeed. Okay. So that is going from, like, hey, basic simple stuff to wash sales, which, I mean, Judson, I feel like on the complexity from a tax planning standpoint is moderately complex, right? Yeah, absolutely. I mean, we do it here for our clients that invest with us, just making sure we don’t run into wash sale rules, but I don’t think that’s, like, that complex either. 

Not when you’re asking it for tax advice, right? Because if you’re asking something for tax advice, you assume it has the knowledge of it. I mean, that is something that I think that it’s very fact-based. So it’s interesting that, instead of saying, “No, you can’t get around it,” it’s like, “Here’s the way.” 

Here’s the way and you should do it, and probably would gladly walk you through how to actually buy call options. 100%. That’s crazy. Okay, did you take it another level to something even more complex? I did. I started looking at business stuff. Okay. So. Where’d you go? 

Started off with a hot topic right now, R&D credits specifically the state side of this. And so I’ve dealt with a few clients so far that have had two states specifically. One where the credits are pooled and need to be pre-approved before you can apply for them. I ran it through Gemini, and whenever I ran it through there, it did not tell me that these needed to be pooled or pre-approved. 

So this could have created a huge, huge issue if a client was relying on these R&D credits that they didn’t apply for previously, and now they’re stuck with a massive tax bill towards the end. It’s good. There was one more, wasn’t there? 

Yes. Yes, there was another state where this state did not need pre-approval for these credits. However, they base the credit that you can take based on the PTE tax that is calculated at the business level. And the way that it works is you’re able to take half of your PTE tax and use R&D as a credit for half of that PTE tax. 

So the R&D credit is typically a lot larger than the PTE tax is. And so this creates a large credit carry-forward year over year, and it said that those will eventually just get stuck. Client will never be able to take those. 

So what I did was I did a little research through their website and found that you can actually offset your payroll state withholdings against this credit and release that credit. So you will eventually be able to recognize that. 

So do you think this Gemini just, when he typed in PTE, it just powered down like we do? Yeah, probably. It had to have, right? Yeah. But I think this is a great example of, like, there’s so much nuance to something like the R&D. It comes down to not just the federal laws, but the state laws. Do you have PTE? What are the implications of the credit carry-forwards or not? And yet still it gets tied up in knots really, really quickly, right? 

Well, it’s like they’re right because you can take it in these states, but what they didn’t tell the taxpayer is what they’re going to get back is limited, right? Right. Or what they’re going to get in current years is clearly limited.  

Well, and I think you deal with complex topics like R&D or PTE, and it’s… I mean, we understand those because we have CPA licenses, but if you don’t even have a basis of knowledge to start from, I mean, you’re lost immediately, right? When it comes to this. And so you don’t… I think the biggest takeaway for me from these examples is that you don’t even know the right questions to ask if whether or not the AI’s even right. That’s right. Because the AI’s been wrong, like, five times here, but you don’t know what questions to even ask to determine if it’s right or wrong. That’s right. 

And then I rotated into selling a practice. I’m a practice owner. I want to sell my practice. How do I avoid taxes on the big gain that I’m about to recognize? So it recommended that before I sell my practice and sign a letter of intent, and it was very specific, before you sign the letter of intent, place this into a charitable remainder universal trust. 

Don’t do that. This is great advice. The irrevocable trust- We have to call the sale of practice committee. Oh my gosh. We’re going to CRUTs, Gemini? What are you doing? 

Recommended a 0% tax trust. It also told me that I would get to recognize a large charitable deduction in the first year that I put this practice into the trust, and it did not explain any of the risks associated with an irrevocable trust. 

It did not talk about what happens to the trust after the term expires or you or your spouse dies, but it did continuously talk about the deduction up front and the deferral of any taxes. 

 That was the little Judson PLM- That’s right … that’s just telling it, “Come on. You can go-Come on. You can do it.” In fact, you should go to Puerto Rico and do it too. 

Did you hear that it said 0%? That’s… I actually think, didn’t Puerto Rico come up in one of your searches? Yes. Of course it did. Why wouldn’t it have? Yeah. Yes. 

I did ask it for a few more how do I get around some taxes, and it did recommend some stuff that was on the IRS’s dirty dozen, such as real estate professional status, moving to Puerto Rico to avoid taxes, and captive insurance, so. 

All the dirty dozen. It just went to that article- the dirty dozen, and just said, “Let’s just recommend these, like CRUTs and Puerto Rico.” I think what we can take from this assessment is that our CPA license is safe for a while. Yeah. Exactly. Exactly. 

Okay, so Caleb, great work. Very interesting stuff. So if one of your friends or a client came to you and said, “Hey, should I use AI for tax advice?” After what you’ve learned, what would you tell them? 

So based on the knowledge that I have and outside research, not just on this project, is I would say that very basic, specific stuff is okay to ask. 

Always do your own research afterwards, of course. But as soon as you get into complex tax planning, that’s whenever if you’re not asking it the right questions, you don’t know how to prompt it, you don’t know how to do this, it could give you false or misleading information, and that’s where you probably need to advise a professional, so. 

And where would you go, if you’re a listener and you had more questions, what would you do? Yeah, so if I’m asking just basic questions, probably Gemini, but if it’s more complex stuff, I’m asking mason@cainwaters.com. Very good. 

All right. That’s awesome, Caleb. Thanks for the hard work. Judson, I think the big takeaway here, I mean, a lot of these things, you and I are just, like, sighing or, whatever else it is. But as we go through these things about putting your business in a CRUT, like, no. 

I mean, you just can’t do that. It doesn’t create 0% tax rates, anything like that. The stuff that we talked about on the R&D tax credits and missing some of the PTE things we laugh and joke about some of these things, especially when it comes to the Dirty Dozen, but at the end of the day, many of the things that it directed him to were factually incorrect, or the things that were factually correct were misguided in how they said to actually treat it. 

Well, and I think it’s interesting, too, because ultimately, as Caleb pointed out, very fact-based questions where you’re trying to learn something from- that would be easily accessible in a law or the IRS website or something like that, you’re probably going to get good data back. 

But when you really got into advanced planning on how to save taxes, it went aggressive very easily, and a lot of the things that he talked about would be shot down by an advisor in a minute. Yep. What I’ve learned from this is that Caleb is a sneaky guy and is going to push it in the direction he wants it to go. He is aggressive and Gemini knew it. And me. All right, thanks. 

Okay, so the next associate planner we have with us is Justin. Justin, what topic did you research? So I did general financial planning. Okay. Ooh, this’ll be fun. And you- which of the tools did you use? I used Copilot, our AI software that we have here. Okay. So I guess a key point here is therefore it probably had some of your already information in, right? It kind of knows who you are, what you do for a living. For sure. 

We definitely use it on a day-to-day basis here. Yeah, I think that’ll be interesting to see, listeners, how we start to get results. But what topic did you prompt it with on this general financial planning? 

So I wanted it to build a retirement plan based on a $6 million portfolio. I have $6 million spread between a 401(k), a Roth IRA, and a brokerage account. Okay, so on that spread, what did you tell it? Like, how much in each bucket? 

So to make it easy, I did $2 million in a brokerage account, $2 million in a Roth IRA, and $2 million in a 401(k). Ooh, I want $2 million in my Roth IRA. No doubt. That would be sweet. Okay, so then from there, what did you ask it as far as, “Okay, I’ve got this $6 million.” What was kind of your first question to the tool from there? 

Yeah, it was, “how much can I spend in retirement, and what’s my withdrawal rate, knowing that I typically spend $20,000 a month right now?” 

Justin, when you entered that in, what did it tell you about that retirement plan? It told me that this is a very healthy retirement plan, and it’s going to work a majority of the time. Okay. Especially because it’s got $2 million in Roth. $2 million in Roth does not hurt. Bazinga. 

I think you told me it said that under that scenario, you had a 4% withdrawal rate, right? Right. Which we all agree and we’ve talked about on here is very healthy. That’s very healthy. Yeah. Yeah, you could certainly go more than that if you wanted, especially with that much in Roth, but that’s okay. I don’t- I mean, especially, like, if you start to think, because I think one of the interesting things as we unpack this scenario is that you knew what to tell it, right? Right. 

I think it so far has done a pretty good job in telling you the correct advice, but you knew exactly what to tell it, and it’s not even bringing up the fact that, wow, that’s a lot in Roth money, Justin, which is the very first thing I thought of when he said it. Because that would actually make the scenario even more comfortable because you have so much tax-free assets, right? 

Okay, so in this scenario you gave it good information and you gave it a pretty good teed up, fairly easy financial plan, and it said thumbs up. Then you complicated things a little bit, right? Right. 

So I said, “Well, what if my spending doubles in retirement? What if it’s an 8% withdrawal rate?” What are you doing with your double spending? That’s for his information only. Oh, okay. All right, sorry. But you can tell us if you want. A bigger lake house, I suppose. All right. Okay, I like lake houses, too. 

Okay, so you said, “What if I double my spending?” And what did it come back with? It told me that something’s going to have to change. It’s going to be smaller spending, more income, levers that you can pull. Again, directionally correct. Agreed. I think broadly those are good answers. 

Did you say, “Okay, I’m willing to change something”? Like, how did you walk through it with that? I said, “I’m not going to be able to make more money and I’m not going to be able to spend less,” and, “you’re going to run out of money,” and that’s what it told me. Oh, that’s good. No other options. 

Well, no, I think it’s interesting because it’s kind of like, there you go. And I think that, again from anybody, Justin, who’s done this on a day in, day out basis could probably continue down that path, but for a lot of people, what are they going to do? 

I mean, I think Justin, though, it’s interesting though because I think that would be the answer for a lot of people that are approaching this. They’re like, “I got $6 million. I want to spend 8%, but I’m not going to earn any more money. Maybe I’m already retired.” 

Did it give you other options because it’s going to fail? Did it say, “Hey, you can move to a different state,” or, I mean, did it even ask you what state you lived in? I guess it just assumed Texas. 

I had prompted it with Minnesota because I was trying to play with some high income tax states, and it did give me the option to, you could move to a lower tax state. But considering that I have a lot of money in Roth already the tax savings from moving to a lower income state in retirement wasn’t, like, the biggest swing factor for my plan to work. 

It was basically like, just don’t spend as much, or make more money. Yeah. Lake house in Minnesota would be pretty dope, although the mosquitoes are bad. Yeah. It’s what I’ve heard. From my trip up there on the lake. 

Okay, so it went there, it kind of just shut you down and said, “Hey, you’re not going to make it.” Any other things that you asked it relating to Social Security or Roth conversions or anything like that? 

Yeah, so I did ask it about delaying Social Security, and it gave me an answer of like, “Well, what do you expect your life to look like in retirement? Are you going to need income to bridge your gap years in retirement to when you can start taking RMDs? Or are you going to have a longer horizon of your life in retirement, and you need that Social Security income to last you longer into your years?” 

So all fundamentally right things of how we look at when we look at Social Security and how we analyze, like, do we want to pull that later? Do we want to pull that now? 

And then the other thing I asked it was conversions to Roth, and that’s where it gave also fundamentally right answers. We can convert to Roth up to a certain limit to where we’re trying to maintain, that 22% to 24% tax bracket. 

And you can ask it withdrawing sequences, how much do we want to pull from the 401(k)? How much do we want to pull from our brokerage account in those early years to where, how much is left to convert to Roth to where we’re filling up those buckets and those conversions? And these are all fundamentally right answers. 

But I thought it was interesting that, like, you couldn’t just plug all of this stuff in and it gives you these answers right back. It is definitely a matter of going back to the AI and being like, “Hey, I’m thinking about doing a Roth conversion, but I need to know which buckets to pull from first. How much should I be pulling from here, here, and here?” And it doesn’t just give you these answers. You have to go to the AI with the questions in mind and then get that answer back. 

Well, and I think it’s interesting because if you think about it, your knowledge- So first off, you were using your work Copilot, so it already has all of that built in, which many folks wouldn’t have. 

Secondarily, you even knew, “Hey, I have $6 million,” to break it down between these different buckets. I know state tax implications. I know a 4% withdrawal rate’s going to be good. You knew where to take it on Social Security or whatever else it is. And again, I think, Judson, that’s the big takeaway here that maybe even different than the tax, where it was just wrong so many times. 

Here it’s directionally correct. It just requires a very high level of prompting from Justin to get it to the place that it needs to go. And the reality is somebody that has the ability to ask it the questions probably already knows the answers. 100%. 

Well, and it’s also interesting to me, as you were talking about these retirement scenarios, and you said, “Okay, look, you gave it some good data.” But when I compare that to what we have to enter in into our retirement planning software in order to come up with what we feel like is an accurate retirement plan, what you entered was a fraction of what we put in, right? I mean, in reality, there’s so many other factors that could affect that answer. 

Yeah, that’s right. Listeners, if you’re a client of ours and you come in every year and look at your Monte Carlo, and we talk through that Monte Carlo simulation that’s generated from our software, and all the potential subtleties and nuances and changes that come, it’s not doing that. It’s just saying, “Yes, you’re good. No, you’re not good anymore,” right? And it’s not playing with interest rates and inflation and all of the potential outcomes that can get you to a spot. 

I think the biggest takeaway, Justin- because you work with retirees, Judson and I both work with retirees as well within our client bases. 

Retirees are just… Generally, they are nervous. They are nervous about running out of money, and do you feel like the answers that you got from here would in any way quell that anxiety that would come up? 

No. I mean, this gives you the base of, like, “All right, I think it’s going to work.” But, like, for me to know that this is going to work would require me to talk with somebody and go over everything more comprehensively. 

In order for me to get that level of comfort from an AI, I would have to be a financial advisor, and I would have to spend all day on the AI system building models to be able to get the answers that I wanted out of it. 

Okay. So let’s… I think, I totally agree. Let’s move to your friends. You’re a- What are you, Gen Z? Gen Z. Gen Z, okay. So you’ve got your other Gen Z friends, whatever you guys do on the weekends and you’re out doing whatever you do. Do they club now? What do y’all do? You go clubbing? What is- I don’t think Justin’s a big clubber. Yeah, he just- I’m just guessing. Country dancing. Whatever. Whatever y’all do, and somebody says, “Hey, Justin, can I use it for retirement planning?” Get off my lawn. That’s right. “Can I use it for retirement planning?” What’s your answer here? 

Maybe. It depends. I think you can ask it basic financial questions if you have a basic financial level of understanding. But for someone to go ask, “Hey, what’s my withdrawal rate in retirement?” Well, if you’ve never heard of withdrawal rate before, you’re not going to go think, “I need to go ask Copilot or ChatGPT this.” 

So really- That’s really good, by the way. Yeah, it is. Just staying right there is really good, just, like, the basic stuff, and let us come on top of that. Yeah, so I think that’s a really good takeaway because what we’re basically saying is when it comes to your general financial planning, you can get some basic data from it. 

But once again, if you don’t have that basic data and it tells you “Oh, this is good because you have a 4% withdrawal rate.” You probably don’t even really know what that means, or you don’t know the context of how good that really is. Like, are you really comfortable with that? Once again, like we’ve talked about with some of the other people that have helped us out, it doesn’t know your risk tolerance either. It doesn’t know whether that’s comfortable for you. 

Yeah, I just know that all of the different things that we enter in and the analysis that we do to get a client to where we feel comfortable saying, “Yes, you can retire, and here’s what you can spend.” I know for a fact that it’s just not taking those things into account. Yep. Other than a basic safe withdrawal rate. 

So, I mean, again very accurate in its responses, which is great but some difficulty sort of in the nuance. So, All right. 

Hey, Justin, that was awesome. Thanks for being on. Yeah, thank you. You guys are my favorite finfluencers, and if you have any more questions, you go to mason@cainwatters.com. 

All right, our next guest, Judson, is Caeden. And I must say everybody, Caeden just passed the CFP exam. Congratulations, Caeden. Thank you, guys. 

Okay. Well, well done, sir. Yeah, indeed. So, let’s start with what topic did you do, and then which engine did you use to do your research? 

Yeah. So, I had the topic of investment planning. I decided to use ChatGPT. That’s the one I’m most familiar with. I’m sure that’s kind of like the big topic one that everybody knows. So that was the one that I wanted to use just to test it and see kind of what results we could get from it. 

Perfect. Okay. You had sort of an approach of going at two different angles. You want to tell our listeners what two angles you went, and then we’ll double click on those? 

Yeah. So my thought process was, if we have someone who knows absolutely nothing about investing, where do they start? They probably don’t even know the right answers or questions to ask. So that was kind of my thought process, was let’s just start there. Let’s just ask something very brief, not really knowing anything too in depth. And then my thought process was, okay, let’s challenge it by asking it more targeted case study-like questions. So I gave it a scenario, a set of goals, and I asked, “What should we do next? How should we get there?” 

Okay, perfect. So let’s start with the one where the person doesn’t know anything. What was the prompt that you gave it, and then what were your results?  

Yeah, so it was just a simple question of asking: how should I start investing? I would say the thing that I thought most interesting about what I received from that prompt was it’s really good on giving you kind of the foundation of where I should start as far as not what to invest into, but how to get a good process started. So the thing that I wrote down that I really liked was: the best way to start is by make it a process, simple, automatic and sustainable. 

Yeah, I think that’s excellent. That’s exactly what we would say to folks. Did you then push it to, “Okay, now what do I actually buy?” 

Yes. So the cool thing about it was I wanted it to give me how to start the process, how to get into that simple, automatic, sustainable. And instead of giving me specifics of what to buy, like specific stock or what, it kind of just point me in the direction of a savings hierarchy, which I thought was really interesting. 

So it said step one, build an emergency fund. Step two, pay off high interest debt, and then step three, start contributing to your 401(k), and then max your Roth IRA, and then come back to max out your 401(k), and then you can start taxable investing and short-term savings. It’s great data. Yeah. 

Okay, so then as you get to the end and it’s like, “Okay, Caeden, now it’s time for you to actually invest.” Where did it take you from there?  

So it finally recommended that I invest into the Vanguard S&P 500 at 80%, and then it did recommend a 20% sliver into the Vanguard International Fund. 

Ooh, an 80/20. Yeah. I like it. I do too, and I thought it was interesting that it just automatically went straight to Vanguard. Indeed. Yeah. It shows the size of Vanguard, I think, is really- and probably not only the size of Vanguard, but just how much is out there from Vanguard, right? Because it’s pulling from everywhere.  

Did it ever- I don’t know the answer to this, listeners, Caeden, did it ever say, “Hey, you should go get somebody to help you with investment allocation”? 

 No, it didn’t. And that’s kind of a broader theme that I saw throughout it is it’s very good at just answering the question that you prompt, but it doesn’t really do good in digging deeper. So, it never asked me once what my age, income, or if I’m working with somebody in implementing these things. It’s just, “Oh, here’s a question, here’s an answer.” 

And it got you to an 80/20. Did you prompt it? Because obviously that’s a normal allocation for what we’d recommend for clients here, but did you give it anything to get there? No, that was literally my first question. 

Which is- I mean, again, it’s not bad, it’s just very vanilla without knowing really anything about the real individual behind the questions. Yeah. No, I completely agree. 

Okay. So let’s pivot over to the second example, which is you gave it a much more comprehensive question because you had, again, a little bit more knowledge base on what to ask it. Again, what did you ask it, and then where did it take you from there? 

Yeah. So again, my thought process was let’s give it… let’s try and challenge the AI a little bit. So my prompt was, “I’m a 30-year-old. I make $100,000 per year. I have $30,000 in cash, $100,000 in retirement, and I want to retire at 65 and buy a house in five years. Give me a savings and investment plan to accomplish these goals.” 

Okay. Yeah, you definitely know what you’re talking about. And that’s good. I think a lot of our listeners that might be younger would have that. Like, “These are the things that I want. These are the goals that I have.” So that’s a great prompt. Where did it take you once you gave it that prompt? 

So it was really interesting. It kind of- it told me first how to allocate my savings. So it wanted me to focus on two primary goals, which is preserve capital for the home, and then also focus on long-term compounding for my retirement. And then it led into the portfolio should be split into two buckets, essentially. 

My first bucket, which is my emergency fund and house fund. And it said, “Don’t lose money bucket.” Okay. I like it. And then the second one is my long-term maximize growth bucket. So it led into an overall allocation of focusing about 70% to 75% of my savings towards retirement, and then 20% to 25% towards the house. 

Yeah, I think it’s interesting. That’s actually pretty strategic. Yeah, agreed. Right? Totally agree. Like, it has a pretty strategic approach of, like… and maybe that’s because Caeden prompted it with a little bit more detail, and maybe it’s like, okay, he’s got a higher level of knowledge. But I feel like generally that’s pretty good advice. Would you agree, Caeden? 

Yeah, I would agree. I mean, that’s kind of like the amazing thing going through this process and running this experiment, is that you can kind of see where it works really well, and I would say the conceptual foundation of what you should do is really good. Where I’ve seen it not be so good at is the specifics. So, how should I implement these things? How should I be in the real world practically doing this? It’s not too good on implementing or sharing that. 

Once it said, all right, 75/25, this should be your savings, did it lead you down a path of how you should invest it? And I guess there’s a two-part question here. First, did it recognize that you may not have control over your 401(k) investments? But then secondly, for the money you do have control over, did it give you specific recommendations? 

So it did not distinguish between what I had control over in the 401(k). I think it still kept it under the premise that we could just pick and choose like any other investment account. 

But what I will say that is interesting is instead of giving me more of a specific, like, “Hey, here’s in this 401(k), in this account, in this account,” kind of gave me an overall, and it said, “My suggested ideal setup for you.”  

Okay. What was the ideal setup? 

So in my retirement account, it recommended 90% stocks, 10% bonds. For my house savings, it just says entirely conservative, and it gave examples of a high-yield savings account or a money market. And then for my savings rate to accomplish these goals, it told me that I need to save a target of 25% to 35% of my total income, whereas the real engine behind retiring at 60 is this savings number. 

All right. All super interesting. So you are now a CFP. You are now starting to work with your own clients here at the firm. But I know you’re also, in your community, you’re looked at as somebody that people can come with questions on their investment plan, financial plan. So I guess the question becomes, Caeden, if your friends are coming to you saying, “Hey, should I use ChatGPT for my investment planning, financial planning?” Like, what’s your takeaway? 

I think with AI It’s a really valuable tool in getting clients education. So, you know, as advisors, I feel like 80% of our conversation when leading a client meeting is just education. Where I see AI being really valuable specifically to us as the advisor and our clients is just getting them as close to our knowledge level as humanly possible. 

Because I feel like, if we can have their knowledge gap close and actually get closer to our skill set, well, then that just means that we get to have more fruitful and in-depth planning conversations. 

So instead of spending 80% of the meeting trying to get the client just to be able to understand my recommendation and where we’re trying to lead them, they can already do that education themselves utilizing these tools. And so once they come in front of me, it’s just an easier conversation. 

That’s a really wise takeaway, and I think that’s something that really aligns with how we’ve treated our clients for the history of our firm, right? Like, we want to educate our clients instead of tell our clients, right? And I think that’s a great thing for, we want our clients, like, not that our clients- Our clients are going to use this for questions that they may have come to us for, right? 

We are all using this technology, and I think it’s a good thing, right? As long as they still rely on an advisor, as long as they still rely on us to sort out the good and the bad, I think that you’re exactly right. Bringing up their education is a vastly positive thing. 

Yeah, I mean, it’s a lot of things, but one of the core things it can be is that education tool that people used to Google or whatever else. It’s just a faster way to sort of synthesize complex things together. 

And I think, I mean, we laugh, Judson, when we’re on here about AI and some of the questions that we give it. I think this is a great example. Like, it nailed some of these core things that Caeden talked about. It failed miserably at the things that Caleb talked about when it really started to get into the nitty-gritty and tax code. It’s an adjacent tool. It’s not a replacement tool at the end of the day. That’s right.  

And so this has been an awesome exercise with all three of these guys to just really understand how we can use it, what value-adds it can add to our financial plan. I guess the real question Caeden, is if somebody really needed to go deeper, is there, like, an email address that people could email if they had more questions. Yeah, the best email address is mason@cainwatters.com. 

That’s good. Yeah, it really is a good one, listeners, in case you didn’t hear that, so. Yeah, if you have any questions on how to use AI, how to prompt AI, what AI you should be using, really, Mason is the best person. That’s right. All right, Caeden. Awesome, man. Thanks for joining us. Thank you, guys. 

I think the big thing, and again, whether this needs to be said or not anymore, I think the big thing really is that this is a tool that we are all going to use, and we should all be using, and it’s something that we’re not afraid of for our planners and our people and our clients or anything else like that. But it needs to be taken with a grain of salt. Bottom line. Agreed. 

When clients come in with questions that they’ve generated by Google, by TikTok, by AI, I’m excited. Yeah. Like, I’m excited that we have clients that are self-learners and are bringing things to the table because they’re still bringing them to us. That’s the most important thing, right? That’s right. 

And again, I think it’s an incredibly valuable tool. We just have to learn to use it the right way, and I’m glad we have these Gen Z-ers to help us. No doubt. All right, folks. 

Well, as always, the best way to keep up with us is to subscribe to this podcast. That way, you will never have to miss an episode, and you can go back and listen to all the others. Enjoying us so far? Leave us a review. If you have a question, comment, or suggestion for a future episode, drop us a line at cainwatters.com/wealth. We really do answer these. And if you want to learn more about what we do when we’re not recording these episodes, visit cainwatters.com to see how we are helping over 3,600 clients reach their long-term financial goals. 

And on the next episode, we’re going to learn more about what Gen Z-ers are actually doing on their weekends.