How To Check AI Financial Advice Before You Act On It

People have always looked for help finding answers, from books and search engines to today’s AI chatbots. Now, more people are turning to ChatGPT, Claude, Grok and other AI tools for instant answers, fact-checking and context, including questions about investing and personal finance. But treating a chatbot as an arbiter of fact can be risky.

Before acting on its financial advice, check its work by verifying the original sources, making sure it reviewed all the relevant information and considering potential bias. My recent experience using AI to analyze an IPO filing showed why those steps matter.

Check The Sources Yourself

In a recent article, I posed a prompt to two LLMs to see what they might flag. I typically use AI systems differently: to check spelling and grammar while I type fast, make a rough transcript of a recorded conversation or help locate a reference when traditional term-based search methods fall short.

However, I always check what I receive, as I did with the article above. It’s necessary to see if the references are correct, which means you need to ask for the links to the references and even page references. These are a minimum.

Make Sure The Chatbot Has The Right Context

I posted a link to my article on X, and someone asked Grok for its opinion on my article via the post. Grok wrote: "Claude called Anthropic’s IPO a buy, citing $16.3B H1 revenue, Q2 operating profit, and $131B cash. It flagged model escapes during tests (admin access, personal data reads), an FTC probe, 5-7% revenue from rule-breakers, and uncancellable compute leases to 2045 as the scariest risk. ChatGPT said no, citing the leases, concentrated control, and unknown price.”

MORE FOR YOU

Don’t assume that a chatbot understood the context of an interaction because chatbots don’t think. They are statistically driven pattern-matching systems.