Employees of large banks say they have problems using ChatGPT to facilitate their work. An oil trader said it gave him outdated information, negating the time saved, reported. Companies like JPMorgan and Goldman have restricted their employees’ use of the popular AI tool. Something is loading.
Thanks for recording!
Access your favorite topics in a personalized feed on the go. download app
Like many others, some financial market traders are turning to ChatGPT to lighten their workload – but they’re running into speed bumps using the artificial intelligence tool, according to a report.
Traders told that they used the popular AI chatbot to help with their work, but found it had several limitations, such as having to spend a lot of time checking facts and deleting outdated information.
ChatGPT could have several applications for businesses in the financial markets, ranging from providing insight into investment strategies to providing stock picks.
But JPMorgan has limited the use of ChatGPT among traders because it fears sharing sensitive financial information with the tool could raise alarm bells with regulators. Other Wall Street banks, Citigroup and Goldman Sachs, have also reportedly blocked use of the tool as part of automatic restrictions on third-party software.
It’s concerns about accuracy, not leaked information, that’s the stumbling block for traders who spoke to . They seem to be using the tool to reduce the more tedious aspects of their jobs, but verifying the information makes some of that redundant.
An oil trader discovered there was outdated information in the crude market outlook written by ChatGPT that needed to be amended, according to . And while a bank clerk found he had compiled a customer overview in less time than an internet search would take, the information he gathered then had to be cross-checked.
“When we talk about high-precision task, it’s worth mentioning that ChatGPT sometimes hallucinates and can generate seemingly convincing answers that are actually wrong,” Morgan Stanley analysts said last week.
“At this point, the best practice is for highly skilled users to spot errors and use generative AI applications as an augmentation to the existing workforce rather than a replacement,” they added.
A January study from Arizona State University also found that ChatGPT was bad at math, with accuracy well below 60% when a professor gave it 1,000 math word problems to solve.
Despite the tool’s potential downsides, Wall Street expects to see a shake-up in AI technology, which already underpins algorithmic trading.
Just over 50% of traders believe AI will be the technology that will have the biggest influence on future trading, according to a recent JPMorgan survey – double the figure from the same survey conducted the year before.