DistantNews
Support us
๐Ÿ‡ฆ๐Ÿ‡บ Australia /Technology

AI Financial Advice: Experts Warn of Inaccuracy, Bias, and Privacy Risks

From ABC Australia · () English

Translated from English, summarized and contextualized by DistantNews.

At a glance

News Named sources Context piece
  • AI tools can provide inaccurate or biased financial advice, according to experts.
  • AI systems may 'hallucinate,' presenting fictitious information as fact, optimized for plausibility rather than accuracy.
  • Users should be cautious about privacy and data security when uploading financial documents to AI platforms.

Using artificial intelligence (AI) for financial advice, such as budgeting or investment guidance, comes with significant risks, experts warn. While AI tools like Microsoft's Copilot, ChatGPT, and Gemini can seem helpful, they are not always reliable.

AI tools can produce inaccurate or biased information.

โ€” MoneySmart websiteThe Australian government's financial advice website on the limitations of AI.

According to Australia's MoneySmart website, AI tools can generate inaccurate or biased information. A key issue is 'AI hallucination,' where the technology fabricates information presented as fact. Antoine Didisheim, a senior lecturer at the University of Melbourne, explains that these systems are optimized for plausibility to keep users engaged, not for accuracy. This can lead users to believe incorrect advice until it's too late, such as a crypto investment becoming worthless.

Professor Uri Gal from the University of Sydney highlights that AI is optimized for plausibility, not accuracy. This means AI assistants might sound convincing but are essentially "following you along." This can be particularly dangerous in finance, where inaccurate advice can lead to substantial losses.

hallucination is one of the reasons these AI systems can provide inaccurate information.

โ€” Antoine DidisheimSenior lecturer in Finance at the University of Melbourne on AI inaccuracies.

Bias is another critical concern. AI systems learn from vast amounts of text, including online forums where financial advice may be dominated by specific demographics. This can lead to biased recommendations, potentially reflecting traditional biases related to gender and race. For instance, AI might offer advice tailored to the risk preferences of young men or avoid recommending risky investments to women.

They're optimised not for accuracy, they're optimised for plausibility.

โ€” Uri GalProfessor of business information systems at the University of Sydney on AI's core function.

Experts also urge caution regarding privacy and data security. Professor Gal advises users to be concerned about uploading sensitive financial documents. He suggests assuming that any information entered into AI tools will be used to train the models, at the very least. The constantly shifting privacy policies of AI companies make it difficult to track, but users should be aware that their data may not remain private.

you chat with [the AI assistant] and it all sounds somewhat sound, but [the assistant is] just following you along and kind of encouraging you along.

โ€” Antoine DidisheimSenior lecturer in Finance at the University of Melbourne on the deceptive nature of AI interactions.
DistantNews Editorial

Originally published by ABC Australia in English. Translated, summarized, and contextualized by our editorial team with added local perspective. Read our editorial standards.