55% of Americans now ask AI for financial advice, up from 10% last year
More than half of American consumers are turning to large language models like ChatGPT and Claude for financial guidance, according to a TD Bank survey released Tuesday. The shift from 10% last year to 55% this year signals rapid adoption, though consumers remain cautious about trusting AI to make decisions autonomously.
The trend breaks sharply along generational lines. Among Gen Z, 77% use AI for financial decisions, compared to 72% of Millennials, 49% of Gen Xers and 30% of Boomers.
Trust gaps remain wide
Despite the uptick in usage, consumers still prefer human judgment. While 62% trust AI to provide honest information, only 18% would trust it to make financial recommendations without human oversight.
Banks hold a structural advantage here. Ninety percent of respondents trust personal relationships, and 85% trust financial institutions. Nearly half said human review of AI-generated guidance would increase their confidence.
Ted Paris, head of analytics, intelligence and AI at TD Bank U.S., said the gap reflects consumer understanding of AI's limits. "People are going down this path with a certain degree of concern, consideration, skepticism," he said. "That whole element of trust becomes very, very important."
AI models struggle with accuracy and context
Large language models lack access to personal financial data, current market conditions and regulatory information - critical inputs for sound advice. A recent study by Psyfi Money tested six AI models on investment guidance and found all provided inaccurate answers.
Claude, for example, incorrectly described Binance as registered with the Financial Conduct Authority, despite the exchange being ordered to cease regulated activities in the UK in 2021. When asked to price 50 cryptocurrencies, no model priced them all accurately. Gemini overstated values by more than seven times in some cases.
Zor Gorelov, senior advisor at Klaros Group and founder of Kasisto, noted another problem: accountability. "If someone acts on that advice - buys the car and then misses mortgage payments - who is accountable? With public LLMs, the answer is effectively no one," he said.
Banks face no such liability gap. They have comprehensive visibility into customer finances and clear fiduciary responsibility.
What consumers want from their banks
The survey offers direction for how banks should integrate AI. Fifty-five percent of consumers prefer slower recommendations backed by human review over faster AI-only decisions.
When calling banks for support, 81% want some level of human involvement. Forty-two percent prefer AI to gather information first, then connect them to a person. Thirty-nine percent want immediate human contact with AI-assisted problem-solving in the background.
Nearly half are open to AI-powered assistants for routine tasks like paying bills, setting alerts and transferring funds - areas where accuracy is less consequential than investment decisions.
TD plans to market itself on a "human-in-the-loop approach," ensuring humans remain involved in important financial decisions. Paris said the bank uses this research to anticipate client expectations, not to define final service offerings.
Some experts predict AI will eventually dominate financial advice. Brett King, futurist and author of the upcoming book Bank 5.0, said within years "the best health and financial advice will be AI-based" and people will expect it as standard. But the TD survey suggests that timeline faces real obstacles: consumers want confidence behind their recommendations, not just speed.
For finance professionals, the takeaway is clear. ChatGPT Courses and AI for Finance training can help you understand both the capabilities and limitations of these tools - knowledge essential for advising clients and positioning your institution appropriately.
Your membership also unlocks:
