You’ve heard about it, you’ve seen it in action, but have you ever wondered what happens when you put Chat GPT to the test? That’s what we explore in this episode of The Agent of Wealth Podcast.
The Bautis Financial team experiments with OpenAI’s Chat GPT, uncovering its potential and discussing its drawbacks. They ask it questions, test its wit, and explain use cases that you may not have considered.
Resources:
Chat GPT | Agent of Wealth Episode 161 – Talking AI With Manuj Aggarwal | Bautis Financial: 8 Hillside Ave, Suite LL1 Montclair, New Jersey 07042 (862) 205-5000 | Schedule an Introductory Call
Disclosure: The transcript below has been lightly edited for clarity and content. It is not a direct transcription of the full conversation, which can be listened to above.
Welcome back to The Agent of Wealth Podcast, this is your host Marc Bautis. On today’s show, I brought on The Bautis Financial team – Kayla, Kyra, and John – to talk about Chat GPT.
Chat GPT is an AI tool that’s generated a lot of buzz over the past year, and the motivation for doing this podcast came back in June when we had guest Manuj Aggarwal on the show for an episode called “Talking AI.”
Related: Agent of Wealth Episode 161 – Talking AI With Manuj Aggarwal
In that episode, Manuj suggested that everyone listening should test out Chat GPT to see how it can help in their daily life. So that’s exactly what we did leading up to this recording.
With that being said, we’ll start off with Kayla. How have you used Chat GPT?
Kayla:
Originally, I was thinking from a financial planning perspective. I wondered if we could use Chat GPT to gather information about a company when a client asks about a particular stock. Basically, I wanted to see if it could provide a quick primer on what the company does and how they make their money. But I ran into some problems here, because Chat GPT only has knowledge up to the year 2021…
Marc:
I had a similar thought… We get asked a lot about AI stocks. Our clients not only want to know about the big AI companies – like Nvidia – but also the up-and-coming ones. So, similarly, I asked Chat GPT to list AI stocks and the response was it only has data up to September 2021.
It’s strange because AI is this cutting-edge technology, yet its knowledge base is delayed.
Kayla:
Yeah, hopefully they find a way to make it more real time. I’m sure they will.
I also decided to ask Chat GPT a really detailed financial planning question involving inheriting an account. There’s a lot of different ways that could impact someone from a tax perspective, so I wanted to see if it would flag those. But, it gave me a wrong answer, basically making up IRS language that doesn’t exist. So, in that case, it can be a bit dangerous.
Marc:
Do you think eventually it will get smarter?
Kayla:
I think it will get better, yes. But it’s hard to say when.
This isn’t Chat GPT-related, per se, but I have Grammarly downloaded on my browser. For those of you who don’t know, Grammarly is similar to spell check. It reviews your emails and work documents, underlining words and phrases that could be improved. When there’s something it sees wrong, it has an AI tool that can help you re-write an email. I tried it out, but you guys would think I was insane if I sent the email that the AI generated. It didn’t sound like a person wrote it at all.
Marc:
It’s interesting how every company is coming out with their own AI product. There’s this mad dash to get the product to market, with less of a focus on it being robust.
It reminds me of when Apple Maps came out. Their top competitor was Google Maps, and Google Maps was far ahead of Apple in terms of accuracy. But Apple decided to launch, saying the technology would learn and become more accurate over time. The company was less concerned about providing correct information, and more concerned about getting people on it. It might be the same thing for AI.
Kyra, you’re up next. What have you used Chat GPT for?
Kyra:
Okay, so I have two examples. You can use Chat GPT to learn a new skill. For example, I wanted to brush up on my Italian before a trip to Italy I’m taking in September.
I asked ChatGPT: “Can you create a study schedule for me to learn Italian in two months?” And it did.
The schedule was broken down into the following sections:
- Weeks 1-2: Building the Basics
- Weeks 3-4: Expanding Vocabulary
- Weeks 5-6: Deepening Understanding
- Weeks 7-8: Refining and Reviewing
Each section had a daily routine for the morning, afternoon and evening, and then a weekend activity. For example, one of the weekend activities was to find a language partner to practice speaking and listening skills.
But, you can also use Chat GPT for this sort of language testing.
I typed in: You’re my Italian teacher, can you test me on some common phrases I should know?
From there, ChatGPT asked me a series of questions in Italian that I then responded to in Italian. You can be specific about the level of your understanding and the amount of questions you want to be asked, as well as the topics you want to be tested on.
Another good use case for Chat GPT is to ask it to summarize books. For example, I asked Chat GPT to summarize the popular personal finance book Think and Grow Rich by Napoleon Hill.
I typed in: “Summarize the book Think and Grow Rich by Napoleon Hill. Give me an outline of the most important concepts.”
The book was organized by its sections – introduction, then sections 1-13, then conclusion – and under each there were two bullet points of takeaways. You get more information from this summarization than you would from the back of the book.
Doing this before reading a book that’s been suggested to you could help you identify if it will be one you’d enjoy.
I also use Chat GPT to help me with small tasks at work.
I might ask for help with:
- Content creation. For example, I’d type in: What are 15 unique personal finance topics I can write about for my company’s blog?
- Writing (or re-writing headlines). For example, I’d type in the topic of a podcast episode and some of the content covered, then ask Chat GPT to give me 10 possible title variations.
Those are just some examples, but overall I’ve found Chat GPT to be very useful.
Marc:
Are those some of the headlines that you’re using for our newsletter?
Kyra:
I don’t use it for the newsletter as much as I use it for blog or video content. But if you are writing a newsletter, you can specifically tell Chat GPT what the content is within the newsletter and ask it to write a subject line and preview text
Marc:
Are you using Chat GPT on the browser? I know there’s a mobile app that’s gotten very popular, too.
Kyra:
I use it on my browser. I haven’t downloaded the mobile app yet.
Marc:
And you mentioned Think and Grow Rich. Did Chat GPT have knowledge of the book already?
Kyra:
Yeah, it has knowledge of everything online, so you can do the same with any book title.
Marc:
Okay. John, what has your experience with Chat GPT been like?
John:
It’s similar to what Kayla and Kyra mentioned. Obviously, accuracy is a concern – you have to be careful what you take from it. But I find it to be quick and useful for certain things… I’ve used it with my kids a lot, similarly to how we use our Alexa device. We’ll ask questions like: How far is the moon from Earth? Silly things like that.
Obviously it has a ways to go when it comes to what Kayla was talking about regarding IRS code. You’re clearly not going to file your taxes based on the advice you get from Chat GPT.
To me, it feels to me like a glorified Google search. When I Google something, I’ll like to look at different resources to feel out how accurate each is. So I’ll look for redundancies, which helps me weed out results that could be right or wrong. But I like knowing the source of the information. Because you don’t get that with Chat GPT, I feel like you have to take the responses with a grain of salt.
Being an avid runner myself, I asked Chat GPT to put together a marathon training program. To be honest, the plan was not bad – I’ve certainly seen worse. There were certainly things that I would change, but the flow and progression was pretty solid. That said, I found similar results on Google.
I tried the same thing with a nutrition plan. As I added more information into the question, like the goals of the nutrition plan, it got way more accurate.
Like everyone has mentioned, there are challenges with keeping it up to date. So it almost feels like the best application for this is towards specific things. For instance, real estate law. You can load every precedent from every case in history towards real estate law, then use it
Marc:
Yeah, like you mentioned, different industries are going to be impacted differently from AI. The legal field will likely benefit, as well as the medical field – whether it’s generating a custom prescription plan or researching different disease diagnoses. Since we’re in the financial planning field, I decided to explore how Chat GPT could help.
I started my experiment by asking Chat GPT to build a budget. I inputted an income and location in New Jersey and it spit back details, such as:
- Spend 25-35% of your income on housing.
- Spend $400-600 on groceries.
- Spend $300-600 on transportation.
It was generic, but still useful. But I wanted to improve the query so instead, I listed out the expenses and asked Chat GPT to create a budget with specific savings goals. For example, I wanted to have an emergency fund, a $10,000 vacation fund, and save in an IRA for retirement. Then, the response was a little bit more useful because it was more specific.
Right now, Chat GPT isn’t able to generate a financial plan, but over time, maybe it will become more advanced.
Then I tried something personal. I found a YouTube video on how to use Chat GPT to draft your fantasy football team. You basically have two options. You can go pick by pick, and when it’s your time to pick, you look at the players that have been drafted, the available players and you make a selection. Or, for a while now, fantasy football has had this concept of an auto draft where it ranks the players and then you don’t have to do anything. It comes to your time to pick and it drafts the next highest ranked player. And this is not very useful because it doesn’t take into account a lot of things like who you have already drafted on your team. Is it a position that you need?
Would it give you too many people and have the same bi weeks? And so this was actually, at least I was very optimistic that this would be something that was useful to use. What I had to do was load roto wire data into Chat GPT to try it out. And then where I had trouble was actually getting the draft data into it. So I couldn’t fully try the experiment, but I guess if I spent more time trying, it would’ve been something that it would’ve been interesting to use because it’s something that I’ve done all the time. I think Kayla, going back to the, and John too, you were talking about stocks and the fact that it’s using stale data. I don’t know how it gets, I don’t know how it’s useful at all for something like that. Like you said, stocks, it’s not even the fact that it’s stale data, stocks price in future data.
So it’s not even close to being ready for prime time when it comes to, in fact, to research. I guess if you wanted to research, I then try to, aside from the AI quote about AI query about what are the best AI stocks I asked it to for dividend paying stocks, I said, give me companies that have raised their dividend rates and I gave it a specific amount each year that have a payout ratio less than a certain amount. And again, it came back with the same thing that it doesn’t have data from later than September, 2021. Now some of that we have the technology ourselves that we can use to research that. So I think like everyone’s mentioned, it’s going to be useful for some things not as useful for others. Of course there’s hype about it. But yeah, I mean I think it sits the unknown of how far it will go in the future and what its use could be.
John:
You hit the nail on the head, Marc. As long as there’s a margin of error, it’s going to be completely useless. If you go to the doctor and they tell you, “We’re 95% sure that what we’re going to tell you today is accurate,” you’re probably going to find another doctor. Accuracy is everything.
Marc:
Yeah. Alright, that just about wraps up today’s episode. In closing, I would make the same recommendation to the listeners that Manuj made to us: Try Chat GPT out. You may find it helpful, it may make you more efficient.
Now, I don’t know how much Chat GPT or AI will help in the finance space right now, so with that being said, if you are seeking help with financial planning, we are happy to help. You can schedule a consultation with our team of financial advisors using the link below.
Thank you to everyone who tuned into today’s episode. Don’t forget to follow The Agent of Wealth on the platform you listen from and leave us a review of the show.
Financial Planning Checklist: What to Consider When Paying Off Student Loans