Artificial intelligence (AI) is seemingly everywhere, and increasingly, seemingly everywhere includes your financial decision-making process.

The good news? There's a place for that.

The bad news? Some people are using AI outside of that place.

And today, we're going to take advantage of the fact that robots haven't yet become sentient to put AI in its place … at least as it pertains to using artificial intelligence tools to manage your finances.

The Tea

Artificial intelligence programs have flooded the consumer realm over the past couple of years. At this point, it's more than likely you've come across at least one of these apps—Siri, Cortana, Copilot, Gemini, ChatGPT, and so on.

Most of these, at the consumer level, exist to spit out writing so you don't have to put undue stress on your gray matter. You tell your AI assistant to make your email a little more sincere, and poof!—it converts "Per my previous email" to "If you'd actually read a f…" well, you get the picture.

Young and the Invested Tip: Want to learn more about AI? Well, these statistics and trends can get you started.

But what good is a technology if we can't apply it everywhere, and sure enough, AI has taken hold in the personal finance industry. Indeed, new research from Experian reveals that nearly half of consumers, most of them 40 or under, are either using or considering using AI-powered tools to manage their finances. And 96% of current users report "positive" experiences with these tools.

The Take

Unfortunately, what tastes great is often terrible for us.

At the risk of sounding like a Luddite … AI can be an efficient, innovative, and even fun resource, but it also comes with significant risks. Those risks include downright inaccurate (while boldly proclaimed) information, and even privacy concerns when apps require a good deal of personal information.

None of this is to say that you should completely avoid AI (though it's worth at least half a thought). But if you're thinking about putting all of your day-to-day financial decisions in the hands of this …

… you should probably keep reading.

That's because this week, we've sat down for a conversation with Grant Gallagher—Head of Wellbeing at New Jersey-based Affinity Federal Credit Union and emergent Weekend Tea regular—about when this technology can truly serve consumers, and when it can steer them astray.

Young and the Invested: Let's start off with the risk that people might be more familiar with—the possibility that AI tools can give inaccurate results/advice. How does that manifest when it comes to financial AI tools?

Grant Gallagher: The source of the information is critical to the quality of the output. 

If you know you're giving an AI tool all of the information it will use, and it's basing its responses purely off of your information, it's probably a reliable response. But so many of these tools are "black boxes." You don't know if they're pulling from other sources, and you don't know what those sources are.

If you knew that data was coming only from financial institutions' websites or Certified Financial Planners®, for instance, that's one thing. But it's going to scrape this information from budgeting forums where people aren't necessarily providing bad advice intentionally, they're just your average consumer that's not as knowledgeable as a financial expert. 

It could literally be pulling info from a conspiracy theory website, or pulling completely unverified information, and somehow finding a way to relate this info to your personal budget. 

You don't want to be the test dummy for the latest hallucination or functionality of AI. And typically, it's not even clear when that's actually the case. So until you have a level of transparency where you fully understand where an AI model is making its assumptions from, you shouldn't be making any serious financial decisions based on what an AI recommends.

Young and the Invested: Does that mean AI is totally unreliable for financial advice?

Gallagher: If you're just looking for basic, generalized information, AI is fine for that. It's probably going to have enough data points and enough information that it has scraped from the web and other sources that it will probably have pretty reliable information. 

There's still the chance of it hallucinating—there might be somebody out there on a Reddit thread saying "You should actually sell all of your assets and go live on an island somewhere." That's completely within the realm of possibility. 

Young and the Invested Tip: "I need a million dollars to work with a financial advisor!" Not true. Here's how much money you actually need.

You're best off looking for very specific, clear-cut examples where AI probably won't give you any information that will potentially hurt you by acting on it. For example, "Tell me ways to save money around the holidays." AI can probably give you some really clear, tangible examples of what you can do there without doing harm.

Young and the Invested: Now, where does privacy risk come in when you're using an AI-powered financial tool?

Gallagher: I don't think people necessarily understand where their information is going when they're putting information into an AI model.

Generally, if you're manually entering in your information, you're probably going to be aware of what you're putting in. It's going to be somewhat limited and censored, probably at a level you're comfortable with sharing. But as tools become more advanced, and you're connecting your financial accounts into tools that are potentially using AI to analyze it and spit out results, you don't know where that information is going, and you don't know the level and depth of what's potentially being shared.

Young and the Invested Tip: Want to DIY your budget? Here's our free template for an easy, basic spreadsheet budget.

One of my biggest concerns when it comes to financial data and AI is the clarity and transparency of who has access to that data, where it's living, and where it's potentially getting shared. I think if we get to a place where [we're seeing more of those tools offered by financial providers], that could be better for the average consumer because typically, financial institutions have much more stringent data security requirements. We have to know all the steps along the way. We have to verify that data is secured, is encrypted, is on lockdown, and it's not going anywhere that it's not [those things], because we have a high bar of regulatory standards. 

But where AI tools are [currently being offered], things are a little too Wild West, if you will, about the data that's plugged into it and where it's going and how it's being used. I would not recommend anybody put any sort of really detailed information into an AI model. 

Young and the Invested: What if you're determined to use an AI tool? What kinds of data, what kinds of privacy risk, are we talking about?

Gallagher: With everything, there's a balance between ease of access/ease of use, and privacy/data security. Your social media, your general info, a lot of that is already out there, whether through marketing channels or, unfortunately, data breaches and consumer leaks. But your financial data is still relatively secure and safe and not regularly accessed.

But there are different levels of financial data. 

There's whether you have loans or not. There's what your credit score looks like. What your balances are. That's super-high-level. Still, some people might not be comfortable sharing that because it will give away some information about their credit quality, their income, their debt situation. But if you're just trying to get some basic advice from an AI model, that's not too much of a concern.

That being said, there are deeper levels of information. At the general transactional level, some people wouldn't be comfortable with you knowing they go to Macy's every week, or a certain supermarket, or a certain adult clothing and novelty shop. If you have an OnlyFans subscription, you're paying for it somehow, that's the reality. Would you be comfortable with someone unknown to you, out there in the cloud, having access to that information? Maybe, maybe not. 

Then there's an even deeper level of specific card transactional data. It's not just your balance and high-level information, it's down to the specific purchase—you went to Walmart and swiped an item that was in this category code. I think the deeper it goes, the less comfortable people will be with this information being shared. 

You have to decide your own level of comfortability with privacy. You have to decide whether it's worth it to give access to that information and have it be unclear where this information is going and whether it might be sold. The deeper and more detailed information you provide to an unknown source, the more you are potentially creating a profile for yourself that can make you more exposed, more easily manipulated, and potentially spend more of your money … on things you might not otherwise buy. KW

Young and the Invested: What would you say is the step, or the area of financial knowledge, where AI just isn't helpful anymore?

Gallagher: I would recommend that people use AI as a starting point—a place where you can just get a broad idea based on your basic information.

After that, though, take it a step further and add a human element: Meet with a trusted financial professional who can look through it and will probably pick up on it pretty quickly if anything doesn't look right. They do this on a regular basis. They have the expertise. They have the knowledge.

When you're meeting with somebody in person, they're going to immediately derive some assumptions off of you. They can pretty quickly tell your age, potentially your background, your level of knowledge and depth on financial topics—something AI is just never even going to think to ask. They might come up with a half-dozen assumptions about you pretty quickly where, unless you think to specifically tell the AI what those things are, they're not going to know to ask those questions, and they won't have those assumptions right in front of them. 

Young and the Invested Tip: If you've ever wondered whether you should get a financial advisor, these questions can help you get a clearer answer.

If you're a layman and you don't know that AI needs to know your long-term ambitions, that you have kids you need to put through college in 15 years, and all sorts of other major milestones along the way, and that you don't have an emergency fund, the AI program is just going to say, "Hey, here's a budget," and that's it. It won't ask you about those important things that you should be factoring in.  

AI also generally isn't regionalized. So if you're in a high-cost-of-living area, it could be basing its decisions off data that's completely different—off national data or data from lower-cost areas.

None of this is a forever statement, but as it stands now, it's still very immature and basic. It's not something I would rely on to give you really in-depth financial advice.

Young and the Invested: On our site, we frequently refer to the idea that, on occasion, the numbers-based decision isn't always the best decision. How does AI handle that?

Gallagher: AI is typically really focused on financially rational decision making. But we're inherently emotional beings who might not be able to execute on a fully rational plan. And that's one of the other benefits that AI still struggles to deliver on: understanding that humans need cheat days, or small wins, or that we sometimes fall off the horse and need to get motivation to get back on and keep going. 

But experts know that, they know how to predict it, they know how to approach people and have proactive solutions to minimize the impact of those issues. 

Riley & Kyle

Young and the Invested

Like what you're reading but not yet a subscriber? Get our weekly financial insights and updates delivered to your inbox every Saturday morning by signing up for The Weekend Tea today! You can also follow us on Flipboard for more great advice and insights.