The FINRA Investor Education Foundation released a report last month called The machines are coming (with personal finance information). Do we trust them?

The self-regulatory organization sought to examine consumer perceptions of financial information generated from AI compared with the same information from a financial professional.

The report revealed that, despite the growing popularity of AI, very few consumers knowingly turn to it for information on personal finances.

“As AI continues to be integrated into consumers’ everyday lives, it is vital to get a better understanding of how they perceive it and how they are using the technology to help make financial decisions, said FINRA Foundation President Gerri Walsh.

“This report found that while more consumers indicated trusting individual financial professionals than AI, there are instances where some consumers preferred AI-generated information related to topics like homeownership and saving,” Walsh said. “These perceptions could change with time, so it will be crucial for the financial services industry to continue to better understand how consumers interact with AI to better equip them with the resources and knowledge to make sound financial decisions.”

Study of 1,000 adults

The report and its findings are based on an experimental study involving more than 1,000 adults in the US who were asked about the trustworthiness of hypothetical AI-generated financial information versus information provided by a financial professional.

The experiment presented four different pieces of hypothetical financial information to a probability-based, nationally representative sample of 1,033 consumers.

“As AI continues to be integrated into consumers’ everyday lives, it is vital to get a better understanding of how they perceive it and how they are using the technology to help make financial decisions.”

Gerri Walsh, President, FINRA Foundation

By design, each of the four hypothetical statements was plausible but not necessarily accurate, and for each question, half of the study participants were told the information was provided by artificial intelligence, and the other half were told the information was provided by a financial professional. Respondents were then asked to what extent they trusted – or distrusted – the information. They were also asked to identify the financial information sources they used and to self-assess their knowledge about personal finance.

The study focused on four topics: home ownership, projected stock and bond performance, portfolio allocation, and savings and debt information.

The key findings

Consumers indicated some willingness to use AI to help manage their finances; but despite some excitement over the possibilities it presents, many remain concerned about AI-generated misinformation.

Key findings from the study include:

• Few respondents reported currently relying on AI for financial advice: Over half of the respondents consulted with financial professionals (63%) and friends and family (56%) for information when making financial decisions, while only 5% indicated they used AI. That said, when it comes to financial apps, which may well incorporate some form of AI, a quarter of respondents (25%) reported using apps for financial information.

• Home ownership information: Respondents broadly trusted information about homeownership, regardless of the source. However, more respondents trusted the information when they were told a financial professional provided it, while more distrusted it when AI was cited as the source.

• Projected stock and bond performance information, Part 1: Whereas a greater share of respondents with lower levels of self-assessed financial knowledge trusted the information when it came from a financial professional rather than from AI (32% compared with 25%, respectively), a larger proportion of respondents with higher levels of knowledge more frequently trusted the information when it came from AI rather than from a financial professional (48% compared with 34%, respectively).
• Projected stock and bond performance information, Part 2: A greater proportion of men trusted the investing information when they were told it came from AI (37%) rather than from a financial professional (27%). Similarly, a larger proportion of white respondents trusted the information when AI was cited as the source (34%) compared with a financial professional (30%).

• Portfolio allocation information: More respondents trusted the information when coming from a financial professional (37%) than from AI (30%).

• Savings and debt information: Respondents generally trusted the information whether it came from AI or a financial professional. However, a greater proportion of black respondents trusted the information when it came from a financial professional (69%) compared with AI (48%).

Other resources

A helpful podcast released in March on the topic of what FINRA is spotting when it comes to member firm use of AI features Brad Ahrens, senior vice president of Advanced Analytics, Andrew McElduff, vice president with Member Supervision’s Risk Monitoring team and Haime Workie, vice president and head of FINRA’s Office of Financial Innovation.

At the FINRA annual conference this May, panelists recommended that brokers and advisers consult FINRA Notice 21-29, Supervision Obligations related to Outsourcing to Third-Party Vendors. The report asks financial professionals to consider what access controls their vendors have, how those vendors’ use of AI could affect their business, and whether LLMs are being used by the vendor to satisfy critical functions related to data integrity or books and records confirmations.

And a June 2020 report from the agency discusses key factors and potential regulatory considerations that securities market participants may want to consider as they develop and adopt AI-based tools.

Implications

Results of this experiment support other research indicating that consumers may not trust AI when it comes to information about personal finances – at least not yet. The perceptions could change as financial technology evolves in the coming years on the AI-positive side, or it could move in the other direction depending on the challenges and errors associated with it.

And the study underscores the importance of keeping a human in the loop in all use cases, as artificial intelligence can help organizations scale up human capacity, not replace it.