Is AI the Answer to Your Money Problems? Here’s What It Can (and Can’t) Do

Some clever budgeters are using chatbots to make better spending decisions. This is what personal finance experts have to say about that.

Full Article

Some clever budgeters are using chatbots to make better spending decisions. Here’s what AI can — and can’t — do for you, according to personal finance experts.

Sahan de Silva wanted more from his budgeting app than it could give him. PocketSmith tracked his monthly cash flow, but what he really needed was help making spending decisions in real time.  “I have this bad habit of going to cafes too much,” de Silva said.  Checking his budget before popping into a cafe could help him keep that habit in check, but he found the experience frustrating. “I had to open PocketSmith, sometimes log in again and then go to my budget page. There was no way that I was gonna do that on the go to check something.” De Silva was already using the artificial intelligence chatbot ChatGPT throughout the day to look up information and ask questions about a variety of topics. So he thought it might be able to help him with this, too. He created a custom ChatGPT thread and gave the generative AI tool his budgeting data using PocketSmith’s application programming interface, a tool that allows one app to “talk” to another. (Finding this API just took a quick Google search.) From there, he could ask ChatGPT questions about his spending, from basic information like whether he could go for lunch that day to post-analysis like which days he tended to spend more on eating out. He could even snap a photo of a pair of sneakers he wanted to buy and ask ChatGPT if he could afford them. The chatbot recognized the image, deduced that it belonged in the “Appearance” category in his budget and then ruled that, yes, he had the money to get the shoes. Pretty awesome, right? For nearly two years now, OpenAI’s ChatGPT has been wowing us with its lightning-quick ability to process what we ask it to do for us: It can write eloquent emails, help plan meals for the week and free up time in our weekly schedules. It also set off a profusion of similar tools, including broad-ranging chatbots like Google’s Gemini, Microsoft’s Copilot and Anthropic’s Claude, as well as more specialized apps. Hardly a month goes by without a leap forward in AI’s abilities.   But generative AI is hardly perfect. The large language models that underlie those chatbots make educated guesses about how to string words together, often with delightful, satisfying results. At the same time, they can whip up responses that may sound convincing but that miss the point or even get things just plain wrong.  So it’s important to approach gen AI tools with care, mixing a healthy skepticism into an appreciation of all the things they can do well — and how they can be truly useful resources. The same applies to any finance software you might use, AI-powered or otherwise.  For de Silva, gen AI does enough right that he finds it helpful in how he manages his money — something many of us struggle with. You don’t need to know what “API” means to benefit from it. Budgeting app developers are finding creative new ways to harness the latest AI technology to provide actionable, easy-to-understand guidance customized to your personal financial situation. With financial education decidedly lacking in the US and a plethora of conflicting money advice out there, this guidance can be invaluable. De Silva likens how he uses ChatGPT with PocketSmith to having a personal CFO to approve his spending or calling a friend to run a decision past them. “It does all the heavy lifting,” he said. AI has plenty of potential to help you make better financial decisions and build your money smarts — without having to wade through a ton of information or rely on a pricey adviser. However, it also has its limitations, and knowing how to avoid them is key to evaluating AI tools and their advice to make sure they’re guiding you in the right direction.    AI may seem like a fairly recent development, and it’s certainly becoming more sophisticated by the day. But it’s been a part of many financial tools since the early 2000s. Artificial intelligence enables computer systems to do tasks that typically require human intelligence. In its basic form, it can gather and analyze large amounts of data, spot patterns and perform automated tasks. If you use a budgeting app, there’s a good chance you’re already using this kind of AI. Popular budgeting apps like Rocket Money and YNAB use AI to do some of the tedious work of budgeting for you. These apps connect with your bank and other financial accounts to import your data, organize it and analyze it using predefined criteria. For example, they can pull in your checking account transactions, sort them into categories like “groceries” and “utilities,” and provide basic insights like how much room is left in your budget and whether you’re on track to reach your savings goals. Their capabilities are often limited beyond that. If you want to dig deeper — say, to figure out why you keep going over budget or whether your savings goals make sense — it’s up to you to analyze your data in the larger context of your personal situation, then scour resources like financial websites, books and podcasts to try to determine which of the conflicting words of wisdom you should follow. I’ve written about personal finance for over a decade, including testing numerous budgeting apps, and even I find this kind of legwork overwhelming and exhausting. But help from AI is on the way. The next generation of budgeting apps — some already released, some in development — employ sophisticated generative AI abilities like machine learning (which enables computers to learn from data without human instruction) and natural language processing (which enables computers to understand and interpret human language) to provide this extra layer of guidance and support. PocketSmith, for instance, has introduced a new feature called AI Tasks. Currently in beta testing, it allows you to interact with ChatGPT from within the app’s interface to ask for insights into your money. The company is looking to determine what people want most from AI, but ultimately, says CTO and co-founder James Wigglesworth, it hopes the app can “deliver immediate, timely personalized financial insights directly to you, like a personal financial assistant.” Here’s a look at some more of these apps, how they’re taking money management to the next level and what you need to know to use them wisely. Gen AI tools such as ChatGPT can do many things, but forming a personal connection with people isn’t usually one of them. And when it comes to something as emotionally fraught as finances, that connection can make a big difference. The budgeting app Cleo aims to differentiate itself by doing just that. Cleo’s chatbot sounds more like a friend than a robot. She — the company refers to Cleo as though she’s a real person — is witty, sassy and pull-no-punches. When you ask her how your spending is doing, she might tell you to “Treat yo’ self, but maybe slow down a bit?” When you ask how to stress less over money, she might begin by responding, “I hear you. This sh*t is real and it’s tough.” Sometimes, she’ll send you a meme. It feels like texting with a brutally honest but loving friend. Part of this personality comes from human writers with a comedy background, who help develop features like Cleo’s “modes” — special chat experiences you can enable to encourage you (Hype Mode), deliver some tough love (Roast Mode) or frame your spending in terms that speak to your favorite fandom (Taylor Swift Mode). But another part comes from Cleo’s LLM, or large language model. LLMs use natural language processing to analyze vast datasets to generate writing and speech that sound like human communication. They’re the engine that enables AI chatbots to hold a logical, text-based conversation with you. Cleo drew from its origins as a Facebook Messenger app when it trained its LLM. By reviewing and analyzing what people asked the app, the LLM was able to hone how it classified and responded to those questions and create customized answers in a smooth, conversational style. Cleo’s personality can make money management more approachable and fun. But it can also inspire a level of trust and comfort that helps people get real about financial struggles they’ve been uncomfortable addressing. “People don’t trust their banks and are disillusioned with the information they might find online, and they feel scared or embarrassed to have a conversation with their friends and family about their financial situation,” said Roxanne Nejad, Cleo’s brand strategy director. “That’s where we feel Cleo is the perfect solution. Cleo is not going to judge you. There’s no judgment here because she’s not a person.” And from the feedback Nejad has seen, people truly appreciate Cleo’s personal touch. “Users have typed into Cleo over 30,000 times, ‘I love you, Cleo,’ and proposed to Cleo over 2,000 times,” Nejad said. I tried it myself out of curiosity. Cleo said she loves me too, but she isn’t ready for that kind of commitment, so we should just focus on my finances. There are basic rules of thumb for smart money management, like spending less than you earn, building savings and paying down debt. But how you apply those rules — and which should you focus on — depends on your unique situation and goals. “There is not a one-size-fits-all approach,” said Jannese Torres, money coach, founder of the Yo Quiero Dinero podcast and CNET Expert Review Board member. “There are so many different things that we have to factor into when we’re creating our own personal finance goals. … Somebody who is a fresh graduate out of college — who is trying to get their first apartment and wants to save up a security deposit — is going to have a very different plan of action than somebody who’s in their mid-30s and has two kids.” Tendi, a budgeting app currently in beta testing, aims to address this need with what it calls “financial personas.” According to John Kalanakis, Tendi’s founder, the app uses machine learning to observe a person’s transactions, identify patterns over time and assign them a financial persona, such as Busy Professional, Frugal Parent or Thrifty Student. “These personas aren’t just labels; they help the platform create recommendations and challenges to best match someone’s specific financial situation and goals,” said Kalanakis. “A ‘Frugal Parent’ might receive suggestions focused on maximizing savings for family expenses, while a ‘Thrifty Student’ might receive advice on managing limited income while balancing education-related costs.” Your persona also allows Tendi to create a customized budget based on your spending patterns. It generates about 30 to 50 potential budgets for your persona, then uses generative AI to see how well these budgets would work for you based on your past transactions and goals. The “winning” plan is the one you’ll see. For instance, your budget recommendations may differ if you want to reduce debt versus if you want to boost your savings. You’re free to tweak your recommended budget however you like, but this starting point could be a game-changer for people who’ve struggled to set a realistic budget in the past. Most budgeting apps require you to set your own spending limits or, if they offer any recommendations, they’re based on how you’ve spent your money in the past — not what’s in your best interest moving forward. In addition, Tendi monitors how well you stick to your budget and where you tend to overspend. This information helps it gauge how well its recommendations serve each persona and adjust these recommendations moving forward. Cleo also monitors your spending and savings behavior to spot areas for improvement, then suggests challenges that can help you do better. For instance, if Cleo notices you always go over your shopping budget, she might suggest you trim your spending this month for the store you visit most often. She then provides several options to do this — like making fewer visits or only spending a certain amount — so you can choose the one that works for you. “We’re really trying to build bespoke features,” Nejad said. “We understand that how people manage their money is different.” A budget that leaves no room for the things you value is often a budget that’s doomed to fail. Knowing what you prioritize and finding a way to afford it can make it easier to stick to your goals without suffering the frugal fatigue that can derail your budget. “One of the things that I always tell my clients is, ‘My budget is not gonna look like your budget.’ I hate cooking, so I budget for DoorDash delivery,” said money coach Rita Soledad Fernández Paulino, founder of Wealth Para Todos and a CNET Expert Review Board member.  “When I see content like, ‘Stop eating out. That’s how you have more money,’ I’m like, ‘No, no one wants to cook three times a day, seven days a week. That’s exhausting. And then the cleanup. No, no, no, no.'” This is something that strictly by-the-numbers apps often overlook. Kalanakis says Tendi is exploring how to address this by allowing people to connect their social media accounts, such as Facebook or Instagram, to the app so it can get a sense of what matters to them. For example, if someone regularly posts about travel, fitness or concerts, Tendi can factor this into the recommendations it gives them. “Once Tendi learns about someone’s passions, the platform makes more relevant financial suggestions,” said Kalanakis. “So, if they are passionate about travel, Tendi might suggest setting up a travel savings fund or recommend ways to cut back in other areas to afford future trips. The idea is to align the financial advice with their passions to produce a better budget that users are more likely to engage with and follow.” The app studies social media posts and uses natural language processing to analyze the content, sentiment analysis to identify whether the emotion behind each post is positive or negative and keyword extraction to identify recurring themes. Tendi is also exploring how image analysis — similar to what de Silva used for his sneaker purchase — can identify themes in the pictures users post. The goal is to create an app that works with you, not against you. “People have passions, and you’ve got to respect those passions and at the same help them achieve their goals,” Kalanakis said. “If someone loves sports and they spend every last dime on sports, you can’t just say, ‘Just stop that.’ We’ve got to help them understand, ‘Well, maybe go to every other game, or maybe we can help find a balance here.’ And that’s where we’re trying to apply AI to really personalize that information.” When we asked AI financial app creators what their mission was in creating their products, boosting financial literacy was a recurring theme. It’s hard to manage your money wisely if you don’t understand the basics. Unfortunately, many of us enter adulthood poorly (if at all) prepared to make the best decisions for our money. Financial education is sorely lacking in the US, with many students receiving little to no education on topics like budgeting, investing and money management. In a 2023 report, the Center for Financial Literacy at Champlain College evaluated each state’s financial literacy curriculum for grades K through 11 and assigned a letter grade from A (financial education course required to graduate) to F (no financial education requirements or offerings). Nearly half (45%) of the states received a C grade or lower. In a recent study by TIAA and the Global Financial Literacy Excellence Center, Americans were only able to correctly answer an average of 48% of questions on financial literacy topics like savings, income, investing and retirement. A quarter (24%) of respondents got seven or fewer of the 28 questions right. Financial literacy is crucial when it comes to successfully managing your money. According to a 2022 report by the FINRA Foundation, respondents with higher financial literacy were more likely to live within their means, have a three-month emergency fund saved up and have a retirement account. They were also more likely to avoid costly credit card fees and alternative borrowing sources like pawn shops and payday loans.  A lack of financial literacy can make it harder to navigate financial decisions and make the one that’s best for your current needs and long-term goals. “[AI apps] are really interesting, shiny new tools that provide really good power,” said Arman Hezarkhani, CEO and founder of the financial app Parthean. “But most people don’t know what question to ask. … A big portion of a financial adviser’s job is to identify the questions that the client doesn’t know to ask and then provide those answers proactively.” Parthean’s AI monitors key indicators of a person’s financial situation, like spending habits and credit card usage, to serve them tips customized to their needs. For instance, if it notices you spend most of your income at the beginning of the month, it may send you a quick budgeting tip, which you can ask the AI to explain further if you want to learn more. Parthean also delivers a daily newsletter with information applicable to your personal finances, like market news about the stocks in your investment portfolio. One of Parthean’s goals is to make financial education more accessible by providing what Parthean’s head of curriculum, Catie Hogan, calls “just-in-time knowledge.” “We’re providing people knowledge about [their] finances as it’s relevant,” Hogan said. “You can talk all day long about Roth IRA and backdoor Roths and this, that and the other thing. But unless it’s specifically relevant to that person at that time, they’re not going to really retain that information or be interested in that information.”  By delivering the insights people need in the moment, in an easily digestible manner, apps like Parthean make it easier to boost your money knowledge and make smart financial decisions without becoming overwhelmed. In a July 2023 study conducted by Qualtronics on behalf of Intuit Credit Karma, nearly half (43%) of respondents said they would be willing to use AI to manage their money if it would solve their money problems. But in a FINRA study released in June this year, only 5% of respondents said they turn to AI to make financial decisions. (Note: This measures whether they knowingly turn to AI. A quarter of respondents said they use financial apps, and some of these may have an AI component they don’t know about.) One of the biggest barriers to AI adoption is distrust. And this distrust is wise. AI is prone to “hallucinations” — making answers up out of thin air, even if the question you ask is clear. These errors can be amusing in some situations. Most people can spot the silliness in AI telling them to put glue on their pizza or eat a rock a day for vitamins and minerals. But when it comes to your hard-earned money, inaccurate or misleading guidance is no laughing matter. And these errors may not be as obvious to spot. Surprisingly, one thing AI isn’t always the best at is numbers. The company behind popular budgeting app Monarch Money learned this when it launched a beta version of its AI Assistant in January. Since Monarch connects to a person’s financial accounts, the company hoped that integrating an AI chatbot into the app would allow people to ask questions pertinent to their own situations. But it discovered that while AI Assistant was able to answer broad financial questions and provide advice, its computational skills were subpar. Those underlying LLMs focus on holding logical conversations rather than running mathematical computations the way a calculator might. “Ironically, it was pretty good at some of the more advanced, nuanced, human-type questions like, ‘Hey, give me some tax advice,’ or ‘What should I consider in this situation?'” said Val Agostino, Monarch co-founder and CEO. “But then when somebody asked it more quantitative questions like, ‘How has my spending on groceries changed over the last three months?’ even though we at Monarch had all that data, we found that it was very inconsistent and often wrong — probably 15 to 20% of the time.” Not wanting to present people with an unreliable product, Monarch decided its AI Assistant wasn’t ready for prime time and disabled it as a default feature on the app. You can still enable this feature in your account settings if you wish, but Monarch has a clear disclaimer that its advice isn’t meant to be a substitute for professional financial advice. Even when the math is right, AI still may not take into account the larger context. When CNET contributor Amanda Smith asked ChatGPT for debt payoff advice, it didn’t seem to fully understand the assignment. It told her to overpay one card by $1,000, didn’t consider that some cards would be paid off before others and recommended payment amounts that pushed her budget into the red. Clarifying her goals with additional prompts didn’t help much. In the end, she wound up moving ChatGPT’s numbers over to Excel and adjusting them the old-fashioned way: manually. Despite the pitfalls and drawbacks, the future is bright when it comes to potential uses for AI in personal finance. Consider George Cockerill, who created a daily podcast about money — his money. This podcast updates him on his spending, identifies areas for improvement and suggests tips to do better. He isn’t a money expert, and he isn’t even a podcaster — he doesn’t write episode scripts, record himself or edit the audio. It’s all done by AI. Cockerill’s process isn’t something the average person is likely to do — he’s a user experience designer who came up with a fun project to teach himself about AI. But recent developments have made it increasingly possible for the average person to create their own podcast using AI. (More on that in a moment.) For those interested in the nitty-gritty, here’s what Cockerill’s process looked like:  He used a combination of ChatGPT and Python code (a programming language) to analyze his budget data from PocketSmith’s API and create a 2-minute podcast script. He then used OpenAI’s API to convert this script to speech and turn it into an MP3 file. Next, he created a custom image for the podcast episode using StabilityAI — a text-to-image generator — and used Python to upload the episode to his RSS feed, which he added to his podcast player. Now he can open his podcast player and get a quick update on his money. The final product was far from perfect. The AI-generated voice was flat, and the tips it gave weren’t always applicable to Cockerill’s personal situation. But what he was able to do speaks to the myriad ways AI could be used in the future to create customized tools that give you the financial information you need in whatever way works best for you. And AI’s capabilities are evolving rapidly. Google’s new podcasting tool Audio Overview is already enabling average people to create their own AI-generated podcasts based on information they provide, without all the technical steps Cockerill took. These podcasts are hosted by two AI hosts who have a conversation about the topic you’ve requested — something Cockerill expressed interest in doing when we spoke. So how can you use AI to manage your money better while avoiding the pitfalls? First, research any AI tool you’re considering. Look up news and reviews of the app, and check out user ratings on the App Store and Google Play to see what people are saying about their experience with it. This can tell you a lot about what the app promises versus what it actually delivers. For example, WallyGPT billed itself as “the world’s first GPT-powered personal finance app,” and it seemed promising when it launched in the summer of 2023. But when I went to review it for CNET’s best budgeting apps list earlier this year, I wasn’t able to register an account after downloading the app. Today, the WallyGPT site isn’t even available. It’s unclear why; we reached out to multiple people at WallyGPT for comment and have not heard back. But clues may lie in its poor App Store and Google Play reviews, which cite issues such as bad transaction categorization, miscalculations and a chatbot prone to errors — including simply not responding to prompts at all. In addition to reading customer feedback, don’t look to AI apps as your sole source of financial guidance. Consider it one tool in your toolbox, but also seek guidance from reputable financial websites, books, coaches or financial advisers to build the foundational understanding you need to vet the AI’s advice. “You need to gut-check the information that you get [from AI tools],” said Natalie Taylor, a certified financial planner, financial technology consultant and company adviser for Monarch Money. “The words they use and the way that it’s phrased sounds like it’s correct, but I think AI sounds just as good giving good advice as it does giving bad advice.” Finally, don’t take a “set it and forget it” approach to AI financial tools. Your needs may change over time as you encounter life events like job loss, marriage or the birth of a child. Consider any AI tool a good starting point, but review your strategy regularly to see if it needs adjusting. The goals you originally set may no longer fit, or you may realize over time that the app’s guidance isn’t giving you the results you hoped it would.  “I never want folks to take their foot off the gas and let stuff go on autopilot because you just never know,” Torres said. “You could discover down the road that this isn’t actually doing what [you] needed to do, but because you put so much faith in it in the beginning, now you have to retrace your steps and figure out what the heck happened. So I say, trust but verify. That’s my relationship with AI.” Ultimately, you’re in the driver’s seat when it comes to your financial success, but AI tools can help make the journey easier. If you’ve been struggling with money management, AI is certainly worth exploring — as long as you keep its limitations in mind. “I think that people should be wary,” Hezarkhani said. “But it shouldn’t keep them from trying.” Visual Designer | Zooey Liao Senior Motion Designer | Jeffrey Hazelwood Creative Director | Viva Tung Video Producer | Mike Viney Video Executive Producer | Andy Altman Project Manager | Danielle Ramirez Director of Content | Jonathan Skillings Editor | Corinne Reichert

Read more at the source