Recent analysis has indicated that widely used artificial intelligence chatbots, including ChatGPT and Copilot, are providing British consumers with inaccurate financial advice.
There are concerns about misleading guidance on topics such as tax and travel insurance, raising questions about the reliability of these technologies in personal finance.
Financial inaccuracies
Tests have shown that chatbots can give poor tax advice and recommend unnecessary services. The technology, which has seen rapid adoption in recent years, is dependent on data sets that may lack the required context for UK-specific financial matters. This can result in advice that is incomplete or ill-suited to an individual's circumstances.
AI financial recommendations are often based on public data, which may not reflect nuanced regulatory environments. Distinctions related to personal tax jurisdictions, home ownership or pension status are not always adequately accounted for, leading to a higher risk of error.
Data foundation
Industry experts emphasise that the effectiveness of AI-driven tools is underpinned by the accuracy and governance of the data used. Many current large language models are trained on information that is widely available, rather than the in-depth, contextualised data needed for sound financial decision-making. This gap can result in users receiving generic or outdated advice.
"AI chatbots are only ever as good as the data and context behind them. Public models are impressive, but they're trained on what's broadly available, not the deeply contextual, well-governed information you need for reliable financial guidance. That's why, right now, these open source tools shouldn't be used as financial advisers. Their answers won't automatically reflect a person's tax jurisdiction or regulatory environment, as well as key factors such as home ownership or pension status. So, the results may be incomplete or inaccurate. However, AI chat bots can still act as helpful access points to further information," said Levent Ergin, Chief Strategist for Climate, Sustainability and Artificial Intelligence, Informatica.
Consumer behaviour
Despite concerns about reliability, consumers are increasingly turning to AI chatbots for financial insights. This trend is expected to grow, making the quality and trustworthiness of these recommendations a greater priority for developers and financial institutions.
Ergin also commented: "Consumers turning to AI to search for financial recommendations is a trend that won't going to reverse. This makes it even more important that, over time, large language models can draw on governed data informed by banks, brokers and insurers. Only then can they surface accurate information, present the right offers and deliver genuinely personalised advice."
Ecosystem challenge
Improving chatbot accuracy will largely depend on creating an ecosystem where data is accurate, governed, and trusted. This may involve greater collaboration between technology providers, financial experts, and regulators to ensure that AI systems are drawing from high-quality and relevant sources.
Ergin said: "Getting this right isn't about the AI alone. It's about the ecosystem around it, building a data foundation that's accurate, governed and trusted."
