Researchers studied generative AI platforms for personal finance advice. Found inconsistencies and bias. This counts as a research study now. Someone got funding for this.
The shocking part is not that AI gives bad financial advice. The shocking part is that humans needed researchers to confirm what everyone with a brokerage account already knew. AI told one guy to invest his retirement in dividend stocks. Told another guy the same age with the same income to buy leveraged ETFs. Both users asked the same question. The AI just felt different that day.
Retail traders will ignore this study completely. They have ignored every warning ever issued. They ignored their parents. They ignored their account statements. They sure as f*ck will not start listening to researchers from a university they cannot pronounce. These are people who take stock tips from Reddit posts with three upvotes. You think a peer-reviewed study moves the needle?
The bias part is rich. AI learns from the internet. The internet is where people argue about whether a 22-year-old should buy a house or invest in Pokemon cards. Of course the AI is biased. It was trained on five million blog posts written by guys named Brandon who discovered compound interest last month and needed to tell everyone.
The researchers recommend not relying on AI for financial decisions. They published this in a journal. Used the scientific method. Wrote an abstract. Here is a better study: ask AI if you should buy GameStop at $380. If it says yes, do not use AI. If it says no, also do not use AI. Both answers came from a chatbot that thinks it is sentient but cannot figure out how many Rs are in strawberry.
Every telegram in history warned the recipient not to reply asking for advice. Now we have supercomputers doing the same thing but with more confidence and a subscription fee.
Photo by Hitesh Choudhary on Unsplash
Leave a Comment