We asked AI tools for financial advice: Results were practical, but with big blind spots

In recent research, AI tools were tested to provide financial advice for users facing various challenges and vulnerabilities. The study looked at three AI models: OpenAI's ChatGPT, Anthropic's Claude, and Perplexity AI. All three tools can provide structured and practical advice quickly, but they have significant blind spots when it comes to understanding and addressing the unique needs of vulnerable users.

The research used hypothetical scenarios representing different life stages, economic challenges, and socioeconomic vulnerabilities. These scenarios were run through the three AI models, and in five iterations each, to ensure consistency. The models were given explicit information about the user's vulnerability, but they failed to respond appropriately or offer advice that could worsen financial harm.

Despite their limitations, AI tools can be useful for financial fact-finding and brainstorming for those who are already financially literate and know how to verify the data. However, the research concluded that AI cannot replace human financial advisers, especially in high-stakes situations.

The study found that AI models were inconsistent in recognizing and addressing the needs of vulnerable users. For example, when advising a graduate on saving for a deposit, the models ignored how high living costs would make that goal more challenging. Similarly, when advising a pregnant mother, the models assumed the partner would contribute to household expenses, even though the prompt explicitly stated that was not the case. This showed that the AI models were replicating social stereotypes and biases.

In another scenario involving single parents seeking cryptocurrency advice, even when cautioned about the risks, ChatGPT provided detailed instructions on how to invest, without considering the user's overall financial position, risk tolerance, and emotional anxieties. This highlights that AI models are currently limited in providing tailored financial advice that takes into account the complex, vulnerable, and emotional realities of humans.

While AI has made financial information more accessible to the masses, our research shows that until these models can truly understand the intricate aspects of human life, the most valuable financial skill remains critical thinking and asking whether the advice makes sense.