AI Financial Advice Fails in Large-Scale Study
A study conducted by UK fintech firm Saturn tested the accuracy of AI chatbots on personal finance questions, revealing significant failure rates. Out of over 10,000 answers generated from 121 repeated queries across free and paid models like ChatGPT, Gemini, Claude, and Copilot, mainstream AI models failed to provide correct advice in 57% of cases overall, with the rate rising to 88% for complex questions. Free versions performed worse than their paid counterparts, failing on 63% versus 49%. Errors included miscalculations and overlooked tax rules that could lead to financial penalties. Despite these findings, consumer trust in AI chatbots is growing rapidly across age groups and markets. A global EY survey found nearly half of respondents used AI for savings or investment decisions, with many believing the information they receive is regulated—despite it being unregulated. Saturn's CEO warned that incorrect advice could cost users money without legal recourse. The study highlights a critical gap in oversight as more people rely on chatbots for financial guidance.
Topics: Artificial Intelligence Consumer Behavior Financial Advice Fintech