AI Doesn't Know Your Client: Why Advisors Still Matter

Mark Hansen 0:00
AI magnifies what's already there. If you put in stupidity, you will magnify stupidity. Okay? If you put in smart, you will magnify smart. But it is just a magnifying glass. The magnifying glass is not the ant is not the bird is not whatever you're looking at. And I think we have to remember we are not trying to take us and outsource us. And if we ever did it successfully, we would, by nature, destroy the original thing. And that's not what we're trying to do.
Seth Harrison 0:35
You are tuned into the Only Human podcast, where financial advisors learn how to stay irreplaceable in a world of AI. Only Human explores strategies for advisors to leverage the unique human edge and thrive alongside technology. I'll be back later with some important announcements. Until then, here's your host, behavioral financial advisor Johans. Enjoy the show.
Yohance Harrison 1:03
Greetings. Welcome back to Only Human, where we talk about being human because I'm human. And my guest, Mark Hanson, he's human. But we both like to use a lot of technology. We are advisors, and we are learning how to adapt this AI technology into what we provide for our clients. And I'm so excited to be with each and every one of you today. My name is Johannes Harrison. You probably know me from the Money Script podcast, maybe from becoming an attending, but this is Only Human, so let's get into it. So I'd like to introduce Mark Hansen. He is the founder of Second Comma, a financial advisory firm. He's also the illustrious host of the Quiet Part. And Mark and I spend some time together with Sonic Boom Coaching, as with Ashley. So, you know, we're going to pretend like we're experts at this podcasting thing because we invested our time, our energy and our money into being better podcasters. So feel free to tell us how we're doing. Mark, welcome to the show.
Mark Hansen 2:00
Hey, thanks for having me, man. It's really good to be here. It's good to see you again in your environment. It's good to go.
Yohance Harrison 2:05
Yes. Let's get it. Let's get it. And you see how he's all perfectly framed. He's got his lighting and his camera, and he sounds good. We both have headphones in. See, we're professionals at this. If you're an advisor and you want to learn more about having a professional podcast, give one of us a ring. We'll happily introduce you to our friends over at Sonic Boom Coaching because they're awesome. I'm just. I just gotta say, they're. They're great. I mean, this is not a commercial for them. It kind of is because he asked me to do that. This is a commercial for them. And it was great. We enjoyed it and we learned a lot, and we learned a lot from each other. That's what we love the most about it. And you heard me say that when I spoke about Ashley's. When Ashley and I were meeting last time. So welcome back to the show or welcome to the show for you, Mark. So, Mark, today we want to talk about AI does not kyc. And for those of you in the industry, you know what KYC is, because we kyc so that we cya. Right. Everybody got it. So KYC doesn't know your client. AI does not know your client. It doesn't matter how much information you feed into your AI system. I don't care if it's Claude or Google. What's Google's called something. Gemini. That's the1. Or ChatGPT. They don't. AI doesn't know your client. It doesn't. And we get to learn how to use AI, but we have to remember that it doesn't know your client. So let's just lay the foundation. First, Mark, tell us how you, as a solo practitioner, are using AI to help scale your practice.
Mark Hansen 3:36
Hmm. I use it in a lot of different ways. Um, and for clarification, in case somebody's starting to boil over already. No, I'm not taking people's personal identif identifiable information and dumping it into chat GPT or something like that. This is more. Hey, I want to. This client has come to me and they have an issue with. Or they're having an internal issue. They're having a family issue. I want to address it. But I also want to talk about A, B and C. How do I introduce this meeting topic? And it just really gives me a starting point. It doesn't give me a final product. But I want to say this. I let ChatGPT or whatever, you know, AI du jour do 80% of the work, and then I put the finishing touches on it. But I'm not using my mental energy to put together that entire thing. That makes sense.
Yohance Harrison 4:24
That makes a lot of sense. So you're using it more as a. As a time saving tool or as a. A what I like to call a writer's block tool, because that one.
Mark Hansen 4:33
Definitely one aspect. I've also used it a lot with, hey, I want to talk to this demographic of people. What are their common struggles? And then it'll come back with, oh, like, I was talking about trying to help business owner families. And it said business owner families are often trying to figure out how to introduce their children to technology so they can adapt more technology into their lives. And I said, no, they're not. They're absolutely not trying to do like technology is self self evident. People like there's entire teams of engineers paid millions of dollars to make it to where you don't have to figure out how to teach your children. Hand a three year old an iPhone, they know how to click around on it. It's pretty amazing. But it, I can go back and forth, I can kind of spar with AI and I can say I don't think that's correct. And you can kind of shave off those edges and it helps me clear out and clarify what I'm trying to say. And then I could say, hey, give me six different ways that you might say this in a way that is not too directional towards giving a specific advice. And I just get to see nice middle of the road stuff. And I can kind of say that is my tone, that's not my tone. And then I can absorb some of that into my meetings and say I do want to talk like that. That is how I want to talk about a 1031 exchange, a reverse 1031 exchange, cost segregation, something like that.
Yohance Harrison 4:35
Oh, yeah.
Yohance Harrison 4:57
Nope.
Yohance Harrison 5:47
Got it. So. So you're not only. I heard you say that you're using it sounds like for some of the upfront marketing conversations you said how to have a conversation with a business owner or family owned business and you're using it with how do I introduce topics to a client, whether it may be a topic you're not as familiar with and how to get those prompters to ask questions to the client to get more from them. And then also a strategy. So here's a strategy that a client wants to implement. What are some things we should be thinking about? What are some ways to maybe go about the implementation of it? Are you using any of the tools that are out there that help with the plan? Writing from an AI perspective with actually.
Mark Hansen 6:34
Delivering the advice, not delivering the advice. That's the piece that I'm not 100% sold on outsourcing to that extent. Now, do I have my general pool of ideas? Yes. Do I have my general pool of ways I like to talk about those ideas? Yes. It maybe in the future, is there a way to systematize that? Maybe. But right now I think it's important that I'm holding that piece. I don't, I don't trust AI enough. I don't think that it knows enough about the actual client to tell them what I want to tell them, how I want to tell them, knowing everything about their situation and knowing that in our previous meeting we just got cut short because they had a family emergency and they didn't even get to finish. They're carrying that burden with them and that's not something that I could just type into Chat GPT and say I want really good financial advice for a business owner who's got 5 million in revenue, has a profitability problem. And in our last meeting they just found out that they've got a multi year series of doctor's visits coming up for their middle child. Chat GPT can't do anything with that. That's me. Does that make sense?
Yohance Harrison 7:37
That makes a lot of sense. And, and even if you, if you wanted to think that Chat GPT could, going back to what you said, they don't know your client, they don't know how to step into that conversation. It's, it's not, it's not people. It's not human. Yes, insert title show here. It's not human. That is why we have to, to your point, there are pieces of the work that we're doing with clients that we have to reserve for our brains and hearts to do the hard work. We can't hand it all off to AI. But you mentioned earlier, something that I want to, to speak on a bit is there are things that you will notice as an advisor that you say over and over and over again to different clients. For instance, when a client says, well, what's a backdoor Roth ira? That is something. I think I can answer the question and get it in a way that sounds like you, that you can copy and paste and say it over and over and send it in an email or in a communic. But I will warn you that don't throw your brain out of the window because I, I like, I like testing AI a lot. I don't know if some of you may remember when AI, this was probably about six months ago when it made this mistake when it was told to spell strawberry and it spelled it with only two Rs instead of three and it knew that it was correct. Most recently there was a gentleman that ask AI to, to provide a 512 digit prime number and it gave one and then it lied about how it came up with the prime number and it kept the lie all the way going no matter how many questions they asked. And the person's like, look, the AI is obviously lying. So he was a mathematician beyond my capabilities whatsoever. I was like, oh, 512 digit numbers got to be prime. Apparently it wasn't. So, yeah, my prime numbers stop at like three. Three digits. 133. I think that's one. Or I could be wrong. The. But AI, at one point I asked a simple question because a client asked me the question, and I said, huh, I wonder. I knew the answer. I was like, I wonder what AI will say? And the question was, can a husband and wife both max out their HSA plan? Now, for those of you at home or in your car or driving to work, whatever it is you're doing, what's the answer to that question? I hope we said no. The answer is no. They both can't max out their hsa. But guess what? Chat GPT said they could. Okay, so imagine if I took. If I believe that AI was correct and I took that information and then gave it to the. To the client. Okay, but think about how many of your clients are out there asking their chat cbt, their Google. What does Google. I keep forgetting Google's name. Gemini. I should know that I was born June. But asking Gemini these questions and getting bad advice, because guess what? AI doesn't know your client. And AI doesn't always know the tax law. So a tip for those of you that are starting to try to introduce AI and using it for information gathering, for prompting your marketing, anytime it gives you an answer, ask it for its sources. Say, hey, that sounds great. What are your sources for this? And especially if it's something that has anything to do with taxes, say, can you point me to the IRS document that you pulled this from? If it can't take you to a website that ends in.gov with IRS in front of it. Grain of salt. Okay, do some more research. Reach out to the cpa. Reach out to the tax attorney, reach out to the enrolled agent. Do some more research. Don't just take it on its face value, because sometimes it'll give you wrong information. Mark, have you ever experienced where AI just got something totally wrong for you?
Mark Hansen 11:33
Oh, yeah. There's been times where AI is a very interesting thing because we do a lot of it in these conversations. The longer your conversation goes, the more confused AI becomes, not the smarter. Yeah, it starts hallucinating. Yeah. So it is very interesting. The more you use it, you're having to actually start over. I was using. I used AI to write an app for my phone. I wanted a custom app, so I used it. I know nothing about coding Android apps, but I've made one, it's on my phone, it's a real thing. But in that process I learned so much about if you talk to it for too long, it can become confused and if you don't consistently give it more framework and more guardrails, it will accidentally fall off the cliff. And it did it all the time. We would delete one function simply to add the next function, which then would break the entire thing. AI has a really hard time tracking itself and it also has outdated information. I think if you read just about ChatGPT it is like somebody who can read the entirety of the Internet. And that sounds so wonderful, that sounds so special, but it is not real time. So something happens today. ChatGPT probably won't even start to intake that information until six more months from now. And so it is never live and we have to remember that. And you get some hallucinations, you get some false data, you get some false assumptions. There was one time I was trying to figure out, we were having a discussion about weather and everybody knows that the quiet part is a secret weather based podcast. But I was trying to figure out what was the largest recorded hailstone ever. And the largest recorded hailstone. I knew this fact outside of AI, it was the size of a cantaloupe. And I thought that's insane. And so I wanted to know how much would that weigh? And So I asked ChatGPT to calculate how much would an ice cube as big as a cantaloupe weigh. And I wish I could find that chat and just show you the number of times. One question, it could not give me the answer. It had all kinds of calculations, but it could not come to a conclusion. And I thought I'm not that great at math, but I could have figured this out by now.
Yohance Harrison 11:46
Hallucinating.
Yohance Harrison 13:38
I. I'm sorry, I was having things flashbacks. We had a hailstorm here in Texas a few weeks ago or I guess about two months ago now. It lasted no more than 15 minutes. Golf ball sized hail. My truck was out in it, 160 dents. Yeah, that was painful. So it's. Yeah. Anyway, so sorry, flashback. It's all fixed now. It's great, you know, nothing a little insurance can't take care of insurance there for when you don't want to pay for stuff that happened.
Mark Hansen 14:12
Insurance, insurance is, is one like a real example. I had a client who family business, a really somebody who did. I didn't want to have too much insurance, but I didn't have too little. But he was Heavily against having insurance. And I did use AI in that instance. And I said, I think this person, generic name John. I think John should have A, B, and C policies, But John doesn't think he should have those policies. John owns X, Y, and Z, and this is why I think he should have these policies. Can you put together some phrasing on how I can gently try to show John that he's currently exposed and that although he is against the idea of insurance, it's actually in his best interest? I did. I have used AI in that kind of an instance. And that was really helpful. Just, hey, in plain English, how would you explain this to. To someone, like, over a cup of coffee or a beer? And that's a really fun way to get ChatGPT to explain stuff back to you.
Yohance Harrison 15:12
Yeah, My favorite is to. I. I typically will use third grader, ninth grader, college student. So I will say, explain this to me as if I'm a third grader. Now explain as if I'm a ninth grader. Now explain as if I'm a college student. And that's how I like to get my phrasing. Because sometimes I need that third grade explanation for not only other people. For me, I'm like, okay, I need this as a third grade level. Because when. Especially when you get documents that you need to review or things that nature like, well, I don't know what they're saying. Explain this as a third grade. It's like, Johnny has a hot dog, Susie has a bun. They need to get together and make lunch, but they're gonna have to cut it in half because there's only one hot dog and one bun. It's like, oh, that's what they're saying. But they'll say that in, like, two pages. I'm like, oh, God, we're talking about a business. It was a. A client sent me a document about merging their business with someone else's business. So they're like, hey, here's the analysis they sent me. Can you make heads or tails? I was like, no, I can't. And I was like, I could throw it in AI though, see what it says. And it came back. I was like, oh. And then, then I was able to say, how did you get to the point of Johnny having one component and the other person having the other component? And it took me to that exact paragraph. And I was like, they could have just used English for that, but they didn't. So that. That can be really helpful. But again, when I went back to the client and Say, okay, now I need to have this conversation with the client. I couldn't just say, hey, AI, what should I tell my client? Because guess what, AI doesn't know my client right now. I could give it some information to say, all right, Susie has been running her business for three years. This is how she feels about it. And I could give it all of that information. But at the same time, I feel like, why that's for me, that's for me to have those conversations with the client. I really want to use AI, like you said, to give me some ideas of how to have the conversation or some of the questions I need to ask, but not to solve the problem for me. Okay. Because it may say that her and her husband can both put maximum contribution to their hsa.
Mark Hansen 15:19
Okay.
Mark Hansen 17:17
I think if you wanted to, if you want to prove this to yourself, if you don't believe us, if you're like, no, no, they just released chat GPT 4.2 Ultra. And it's amazing. It's. It's different, Johans, it's different. I would say use your own information because we don't want to take clients information and put it in there, but just start feeding real information in there and try to get ChatGPT to do all of your job. And you will, you will see that it's actually way more work. As in, hey, help, I'm trying to take Mark. I want to explain to Mark that he needs, call it $3.5 million of term life insurance. How do I help Mark understand that he needs this? And it'll come back with something generic and you'll say, well, no, Mark's life is a little different. It looks like this, it looks like that. And then ChatGPT will update what it says and then you say, well, that's almost it. But. And you'll go back and forth, 10:15, when you get to the 20th time, you just need to realize you could have explained this by now. You've explained it to the computer. Could have just said it to the human.
Yohance Harrison 18:11
Thank you.
Yohance Harrison 18:14
Thank you. And that's okay. That's also part of being human is to show your own vulnerabilities, to show your own concerns and to have the conversation. Because that's especially when it's something with, with insurance. A lot of times with the insurance, it requires the conversation. Because if everyone just purchased any type, and we're not just talking about life insurance, we're talking about health insurance, disability insurance, auto insurance, home insurance. If everybody just purchased the right amounts and types of the insurance, that they needed, just based on facts, we wouldn't need insurance salespeople. Okay? We wouldn't. And as a matter of fact, I'm willing to bet the majority of your clients, unless this is something you focus on, their home is dramatically underinsured.
Yohance Harrison 19:00
Okay. And they could have gone on esure.com and use their AI and said, oh, it looks like you need $750,000 worth of home insurance or to rebuild their home. Mind you, they're sitting in a home that's worth 1.5 million. Yeah, you're not going to rebuild that. Okay. But most, most home insurance policies dramatically underinsure the value of the home at first glance. And most people don't go back and update it. Home values have gone up what, 20, 30% in the last five years. How many people have updated their cost to rebuild in their policy? Okay. So again, that's, that's where we as the, as humans, when it comes to those insurance, especially insurance conversations, they need us to have the conversation with them to figure out what their hang ups are, to figure out why, you know, why they're so afraid to separate themselves from the premium in order to get the coverage that they need. The same is true as we were going through, which I know it's calmed down by the time you all hear this. Maybe it'll be even calmer or it would be worse. I don't know. This whole tariff thing. The market went bananas for like two weeks on the tariffs and now it's quietly just calmed down. And as Mark was saying, AI really? I tried to even talk to AI during. I was like, oh, you, you guys haven't even caught up. And I had to tell them go read this recent Wall Street Journal article. And they're like, oh, really? I mean, AI was shocked too. It's like, really? That just happened? Like, yes, AI happen.
Yohance Harrison 20:25
Like, is this a movie plot? It will be soon. But, but, but what our clients need is the conversation. I think it was, it was Carl, the artist, Carl Richards said best on one of his, one of his paintings. He said it was, it showed what I need when the market's going crazy. And it was, it was a bar chart and it had facts and hugs and hugs was bigger than facts. Okay. And that's in the hug. That's that emotional component that we bring to the table that our clients need, especially in their time of need, especially in a time where things are very volatile. So, so any closing thoughts there, Mark? On, on AI not knowing your client, that you know your client and how to use it as a better tool.
Mark Hansen 21:20
Yes. I would say for anybody who feels like they're behind the curve on AI because they don't know enough or because they haven't utilized AI enough, that. That's not. I don't think that's the goal. The goal isn't to take Mark and recreate him as an AI because then you actually, if you were to successfully do that, you would also, by nature, destroy Mark. It's. I think it's important that we understand these are two totally separate things. One, AI, AI can help Mark get to a solution faster. AI can help Mark think of it in three different ways simultaneously. AI AI magnifies what's already there. If you put in stupidity, you will magnify stupidity. Okay. If you put in smart, you will magnify smart. But it is just a magnifying glass. The magnifying glass is not the ant is not the bird is not whatever you're looking at. And I think we have to remember we are not trying to take us and outsource us. And if we ever did it successfully, we would by nature destroy the original thing. And that's not what we're trying to do. Hopefully some of that sinks in somewhere.
Yohance Harrison 22:27
Talk about a hot take. Thank you, Mark. All right, so here's the fun part of the show. Now, we had some changes in personnel here at Manuscript. Well, so as we were preparing for this show, have to admit, Mark didn't get all his preparation tools, which is. Okay, that's not like. But it was. It was easier because I was like, oh, it's Mark. He'll be good. He's a professional because he went through sonic boom coaching and he did great. So here's the part of the show that you may not be as familiar with. So, Mark, we like to play a game here on Only Human, where we talk about the technology that we use as advisors. And we're going to put the technology, three pieces of technology into three different categories. Marry, divorce, and maybe date on the side. Okay. So what technology are you absolutely married to and will stay married to forever to death? Do you part. What pizza or application? Technology. What are you. What. What technology or application. Excuse me. Software. Think software. Like apps that we use. Do you absolutely want to divorce? And what is up and coming that's starting to get your eye? You know, like the. The little meme where the guy's walking with his girlfriend, he's turning, looking behind him.
Mark Hansen 22:28
My goodness.
Mark Hansen 23:15
Okay.
Mark Hansen 23:37
Okay, let's say the one on the side that's airtable. I'm I'm starting.
Yohance Harrison 23:44
I don't know a lot about Airtable. What is Airtable?
Mark Hansen 23:46
Airtable is just like a spreadsheet on steroids. It's think of just two dimensional storage but you can. It's got like a. It's like building a third dimension. Airtable has the third dimension where you can store different types of variables inside of the spreadsheet. It scales up and down very easily and you can use it in a lot of these AI make.com innate in situations. You can embed web links and all kinds of stuff in there. It's got that third dimension and that's to maybe lead into the one that I'm currently married to Microsoft Excel. My love language is Microsoft Excel. But I'm definitely starting to get some side eyes for Airtable and I don't know what's going to happen there but I'm interested in seeing what would I divorce.
Mark Hansen 24:35
I'm trying to think. I think okay, if pretty much any. I think I'd be totally happy getting rid of all in any social media. I spent a couple of years doing that and then got into this, this line of work and I was like actually I need a social media presence because I want to be able to reach people on platforms. But I think I could be okay without social media. I like networking face to face. I like the. The only human aspect of networking. I'd be okay leaning into that and divorcing social media.
Yohance Harrison 25:03
Got it, Got it. Beautiful. Mark, thank you so much for sharing with us on Only Human. Please do us a favor. Tell everyone how they can find you.
Mark Hansen 25:14
You can find me on all the major platforms but the one that's easiest to get a hold of me is going to be LinkedIn. So just search for Mark Hansen. That's H A N S E N. Not like Hanson, the boy band from the 90s.
Yohance Harrison 25:25
Hanson, like the soda. There you go.
Mark Hansen 25:27
Yes.
Yohance Harrison 25:29
Oh man. All these new sodas popping up the Oli or O Boly and they're like where's good old fashioned Hanson's? But yeah, I was a. Those are finer days back then. So do yourself a favor, share this with someone else. How does that do a favor for you? Because now you can have a conversation with them about what you learned and you can try different things out and you can bounce ideas off of each other about A.I. do yourself a second favor, play around with A.I. a little bit. Okay. It's. You're going to have to get used to it. It is today's electricity. Okay. It's going to be a part of just about everything that we do. It just has the fancy buzz name right now, AI, but it's going to be a part of everything that we do soon. So it's one of those things where it's like you're going to have to get on the train eventually. If you don't, you will get left behind. As we said, it's not going to replace you, but those that know how to use it will. So until next time, we'll see you on Only Human Podcast. Please subscribe. Like Heart Share. Do all those cool things that we do on social media and we'll see you next time.
Mark Hansen 26:35
Take care.
Seth Harrison 26:44
Hi, I'm back. Did you enjoy being a fly on the wall for that conversation? I know I did. Now remember, don't try everything you learned today. Pick one or two things you can implement and implement them well. And don't just try something for a week and give up when it's not working. Successful implementation takes time, energy and repetition. Take a moment to follow today's guests and remember to consult your compliance professional. We don't want you to get into trouble. Do you have a group of advisors who want to improve their human connection in the AI world? Contact our team to have Johan speak at your conference or event before you go. Please, like comment and subscribe. If you're on YouTube, click the bell for notifications.