19% lost $100 or more on bad AI financial advice

A significant proportion of people are following financial advice given to them by AI models, and many reported losing $100 or more from the advice. 

That’s according to a recent survey from Pearl, an AI professional service platform, which found, among other things, that 20% of people have acted on financial advice given to them by AI, and 19% of people reported losing $100 or more due to following this advice, a number that jumps to 27% when considering only those from Gen Z. A spokesperson from Pearl, in an email, said it was up to the individual respondent to determine what constituted “bad advice.” The poll did not ask how they lost this money (e.g. fines and penalties, bad purchases, etc.) 

As to what kind of financial advice they are following, the poll found that 9% have acted on tax advice given to them by AI, 22% say they use AI for stock-buying advice, and 21% said they’re buying cryptocurrencies based on AI suggestions. A significant proportion of respondents, 27%, believe AI can give them all the financial advice they’d ever need.

This is part of a wider trend of people turning to AI for professional services normally provided by human beings. Over the past six months, 65% of respondents said they acted on advice they got from a chatbot that they previously only trusted from a human professional. Beyond the 20% who have made financial decisions based on AI, 23% have turned to AI for tech support, 21% have made medical or veterinary decisions based on AI advice, and 19% followed home improvement advice from AI. The poll found that many would likely follow legal advice as well, considering 31% would let an AI lawyer defend them in court if it meant zero legal fees, and 28% would sign a legal document drafted entirely by AI. 

There are several reasons why people are turning to AI for sensitive advice like this. For one, professional services are seen as too expensive, so even if they wanted to talk to a human accountant about something, they believe they would not be able to afford it. Another factor is lack of access: one in three, for example, said they could not reach emergency medical services within 15 minutes. Another is declining trust in experts overall. For medical professionals specifically, 22% trust social media influencers’ advice more than doctors’ advice, 23% trust AI medical advice more than doctors, and 28% trust their family’s medical advice more than doctors. We also see more people turning to social media for tax advice, as the IRS recently hit taxpayers with over $162 million in penalties for claiming fraudulent tax credits they heard about through social media (see previous story.)

This is despite the fact that people generally recognize that AI can make mistakes, sometimes significant ones. Only about half (54%) find current GenAI solutions like ChatGPT to be trustworthy, down from 66% in December. And yet, 61% of respondents have actively followed advice from an AI tool in the past 30 days in at least one area of professional expertise (personal finance, medical legal, etc.) What’s more, 29% of respondents agreed they rarely double-check the advice given by AI. In other words, nearly a third are not verifying AI outputs with a second source.

They are paying a price for this: in addition to those who’ve lost money following AI generated financial advice, the poll also found that 22% of the respondents have followed medical advice from AI that later was proven wrong. Andy Kurtzig, founder and CEO of Pearl.com, found this to be an alarming development.  

“We’re seeing a dangerous trend of misplaced trust born from desperation,” he said. “Our data shows 31% of Americans would let an AI defend them in court to avoid legal fees, a shocking statistic when Stanford research shows AIs can be wrong about basic legal facts 60-80% of the time. This isn’t just about legal issues; 23% of respondents now trust an AI’s medical advice more than a doctor. AI lacks the nuanced judgment, ethical training, and real-world experience that are essential for high-stakes situations. An algorithm can inform, but only a human professional adds the required wisdom.”