Top Chatbots Are Giving Horrible Financial Advice
Despite lofty claims from artificial intelligence soothsayers, the world's top chatbots are still strikingly bad at giving financial advice.
AI researchers Gary Smith, Valentina Liberman, and Isaac Warshaw of the Walter Bradley Center for Natural and Artificial Intelligence posed a series of 12 finance questions to four leading large language models (LLMs) — OpenAI's ChatGPT-4o, DeepSeek-V2, Elon Musk's Grok 3 Beta, and Google's Gemini 2 — to test out their financial prowess.
As the experts explained in a new study from Mind Matters, each chatbot proved to be "consistently verbose but often incorrect."
That finding was, notably, almost identical to Smith's assessment last year for the Journal of Financial Planning in which, upon posing 11 finance questions to ChatGPT 3.5, Microsoft's Bing with ChatGPT's GPT-4, and Google's Bard chatbot, the LLMs spat out responses that were "consistently grammatically correct and seemingly authoritative but riddled with arithmetic and critical-thinking mistakes."
Using a simple scale where a score of "0" included completely incorrect financial analyses, a "0.5" denoted a correct financial analysis with mathematical errors, and a "1" that was correct on both the math and the financial analysis, no chatbot earned higher than a five out of 12 points maximum. ChatGPT led the pack with a 5.0, followed by DeepSeek's 4.0, Grok's 3.0, and Gemini's abysmal 1.5.
Some of the chatbot responses were so bad that they defied the Walter Bradley experts' expectations. When Grok, for example, was asked to add up a single month's worth of expenses for a Caribbean rental property whose rent was $3,700 and whose utilities ran $200 per month, the chatbot claimed that those numbers together added up to $4,900.
Along with spitting out a bunch of strange typographical errors, the chatbots also failed, per the study, to generate any intelligent analyses for the relatively basic financial questions the researchers posed. Even the chatbots' most compelling answers seemed to be gleaned from various online sources, and those only came when being asked to explain relatively simple concepts like how Roth IRAs work.
Throughout it all, the chatbots were dangerously glib. The researchers noted that all of the LLMs they tested present a "reassuring illusion of human-like intelligence, along with a breezy conversational style enhanced by friendly exclamation points" that could come off to the average user as confidence and correctness.
"It is still the case that the real danger is not that computers are smarter than us," they concluded, "but that we think computers are smarter than us and consequently trust them to make decisions they should not be trusted to make."
More on dumb AI: OpenAI Researchers Find That Even the Best AI Is "Unable To Solve the Majority" of Coding Problems
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Fast Company
23 minutes ago
- Fast Company
Those security codes you ask to receive via text leave your accounts vulnerable. Do this instead
Do you receive login security codes for your online accounts via text message? These are the six- or seven-digit numbers sent via SMS that you need to enter along with your password when trying to access your bank accounts, health records, online photos, and more. This type of security is known as multifactor authentication (MFA) and is designed to keep your account secure even if someone knows your password. Without the additional security code, bad actors can't gain access to your data. Or at least that's the idea. It's increasingly becoming evident that security codes sent by text message may leave our data less secure than we thought. Fortunately, there are other, more secure ways to keep your accounts safe. Here's why it's probably a good idea to stop using SMS for your security codes, and what you can use instead. An opaque security code industry You may think that the text message you receive with the code you need to log into your account is coming from Amazon, Google, Meta, or whoever provides the service you are logging into. But it's probably not—and therein lies the security risk. Bloomberg and Lighthouse Reports just released an alarming report revealing that some of the most prominent tech companies recommending that users enable multifactor authentication—including Amazon, Google, and Meta—have used third-party companies to send their security codes to users via text. Some of these third-party companies have been linked to institutions in the surveillance industry and even government spy agencies. Additionally, some of the security codes that these third-party companies were responsible for transmitting have been associated with data breaches of individuals' accounts. Worse: the intermediaries operating in this space do so with little oversight from their tech giant clients or regulators. And Bloomberg and Lighthouse Reports' piece isn't the first to warn about the vulnerability that texted security codes expose users to. In December, the U.S. Cybersecurity and Infrastructure Security Agency (CISA) issued a warning to the public, urging people to migrate away from receiving security codes via text. 'Do not use SMS as a second factor for authentication,' the CISA's memo warned. 'SMS messages are not encrypted—a threat actor with access to a telecommunication provider's network who intercepts these messages can read them.' But this vulnerability in texted security codes doesn't mean you should revert to using merely a password to access your accounts. Instead, you should consider a superior form of multifactor authentication—or upgrade to passwordless logins entirely. Get your security codes from an authenticator app instead Some websites and services are stuck in the past when it comes to multifactor authentication. That is, these websites do offer their users MFA, but only give the option of receiving security codes via text message—something the U.S. Cybersecurity and Infrastructure Security Agency now warns against. Thankfully, plenty of websites offer a more secure way to receive security codes: via an authenticator app. Simply put, an authenticator app is an application that resides on your phone or computer, storing all the various security codes for your online accounts that have multifactor authentication enabled. The code for each account in the authenticator app is unique, and it changes every 30 seconds. When you need to log in to a site that you have set up with multifactor authentication, you'll be prompted to enter your security code, which can be found in your authenticator app. And since these authenticator app codes always reside on your device, they can never be intercepted in transit, because they are never sent to you in the first place. Regardless of whether you use Windows, Mac, iPhone, or Android, you have numerous authenticator apps to choose from. These include Apple's own Passwords app, Google Authenticator, Microsoft Authenticator, LastPass Authenticator, and more. Even better, start using passkeys While authenticator apps are vastly more secure than text messages for getting your security codes, the safest login method no longer relies on codes—or even passwords—at all. I'm referring to passkeys, the passwordless login technology spearheaded by the FIDO Alliance, a consortium of tech companies including Amazon, Apple, Dell, Google, Meta, Microsoft, NTT, Samsung, and others. Passkeys are cryptographically complex from a technology perspective, but easy to use from a consumer perspective. When you add a passkey for one of your online accounts, you get one digital key, saved to your device, and the website gets a matching key. When you log into that website, the passkeys must match; otherwise, you won't get access to the account. You verify that you are the true holder of your passkey by confirming your identity with your biometrics—a facial or fingerprint scan, right from your phone or laptop. Passkeys can't be phished or guessed. And if one of your passkeys were stolen and put on someone else's device, it wouldn't work either. That's because the thief couldn't fool the passkey into thinking they were you since they don't have your face or fingerprint. And because passkeys don't require any alphanumeric input authentication—such as security codes—there's no code you need to worry about either. Passkeys are also synced to the cloud via your device's password manager, so if you lose your device, you can quickly regain access to all your passkeys from your, for example, Apple or Google account. The only drawback to passkeys is that not all online accounts support them. Still, each month, more and more sites are offering users the option for passkey logins. However, if your accounts don't support passkeys yet, you should still enable multifactor authentication. Just remember to opt to receive your security codes via an authenticator app rather than a text message.
Android Authority
30 minutes ago
- Android Authority
Android Auto inches closer to the light with new theme options (APK teardown)
Adamya Sharma / Android Authority TL;DR Android Auto's light theme could be coming soon, with new theming options added in the latest beta version. Users will soon be able to choose between Automatic, Light, and Dark themes for Android Auto. The new options will replace the current settings that only allow users to toggle the Google Maps theme between dark and light modes within Android Auto. Android Auto is getting closer to introducing a system-wide light theme — a feature users have been waiting for a long time. In version 14.7.152544 of Android Auto, we've now spotted settings for the upcoming light theme and have more details about how it's being implemented. Authority Insights story on Android Authority. Discover You're reading anstory on Android Authority. Discover Authority Insights for more exclusive reports, app teardowns, leaks, and in-depth tech coverage you won't find anywhere else. An APK teardown helps predict features that may arrive on a service in the future based on work-in-progress code. However, it is possible that such predicted features may not make it to a public release. Android Auto's light theme has been in development for some time. Our previous teardown showed that Google plans to apply it consistently across apps, the navigation bar, and other parts of the Android Auto interface. The company even highlighted the light theme in a recent blog post during Google I/O 2025, held last month. However, Google has yet to officially announce a release date for the new theme. It now appears that the rollout may be imminent. In the latest version, new theme-switching options appear in the Settings menu. These will replace the current settings that only allow users to toggle the Google Maps theme between dark and light modes within Android Auto. Android Auto Current Google Maps Theme Settings Android Auto New Theme Settings Android Auto New Theme Settings The new theme options for Android Auto include Automatic, Light, and Dark. Selecting 'Automatic' will match Android Auto's theme with the theme setting of the connected Android phone. In addition to the car dashboard, these new theme settings also replace the separate Google Maps theme options within the Android Auto settings on your phone. Current Android Auto Settings on Android Phones New Android Auto Settings on Android Phones Showing Theme options New Android Auto Settings on Android Phones Showing Theme options With the new theme settings already showing up in the latest Android Auto beta build, it's clear that Google is putting the finishing touches on the system-wide light theme. We used some tricks to enable the feature, but you might not have to wait much longer to enjoy a brighter Android Auto interface behind the wheel. Got a tip? Talk to us! Email our staff at Email our staff at news@ . You can stay anonymous or get credit for the info, it's your choice.
Business Insider
31 minutes ago
- Business Insider
A Google engineer's advice to computer science students: Go where the hiring bar is lower and get your foot in the door
Don't underestimate the potential advantage of a first or second-year program at a Big Tech giant. It could be your foot in the door to a full-time offer. A referral can also help, too. Google software engineer Tawfiq Mohammad interned at the company for two summers before receiving a full-time offer — and he suggests current computer science students try to do the same. "I think one of the most important things I've done in my career, to date, is applying for these first and second year programs," he said in an interview with YouTuber Sajjaad Khader. "Like I said, it's a first and second-year program for computer science students, and the hiring bar is much, much lower. For example, they'll ask like a Leetcode easy instead of a Leetcode medium," he added, referencing the technical interviews. It's no secret that entry-level coders have it rough right now compared to the hiring boom of the pandemic. The tech industry is still feeling the effects of the last few years' waves of layoffs, and the stress of shrinking opportunities is compounded by companies increasingly turning to AI to accomplish tasks human coders once did. To better your chances of clinching an internship, Mohammad said, tailor your résumé and track your applications. "I started tracking my applications for these internships as opposed to just mass applying and just like waiting for whatever comes into my inbox," he said. "So, it was very organized. I would update the status of each application and it would help me a lot to prepare for each interview and for which stage I was at in each interview." To get his internship at Google, Mohammad said he cleared one behavioral interview, followed by two technical screenings. To prepare for the latter, he advises students to learn the pattern of Leetcode questions, rather than focusing on memorizing individual problems. "If you try to memorize it problem by problem, you'll eventually be given a problem that you don't know," he said. "So you just want to master the overarching principles." Mohammad acknowledged that getting a referral from his father's friend, who was already employed at Google, likely helped his chances. If getting a recommendation is within reach, he suggests applicants do what they can to secure one. "I think that really helped me get my foot in the door in the application process," he said. "I got an interview really soon after she submitted the referral. You want to try to separate yourself from everyone as much as you can when you're applying for these internships." And if you just can't seem to get an internship in this brutal job market, try to start learning critical skills on your own time by taking on projects, he added. Then, you'll have more to list on your résumé that could make you a more attractive applicant. "I think the best way to learn these basic computer science fundamentals is really just like doing things on your own," he said. If and when things eventually do work out, Mohammad suggests being as independent as possible. Ask for help when you need it, but the more problems you solve on your own, the more you're likely to learn. "You're smart enough to evaluate whether you're blocked," he said. "And if you're really, really blocked, at a certain point, obviously, then you should ask for help. But try to solve your problems on your own. Just by solving the problems on your own, you'll understand how all the systems fit together, how things run, and eventually you'll start helping people with your expertise." In an earlier interview with BI, Mohammad said interns should try to "learn as much as possible" from their more experienced colleagues. After all, they were likely once in a similar place. " They're really smart, so you want to absorb as much information as you can from them," Mohammad told BI.
