Competing LLMs Were Asked to Pick Stocks. Their Choices Revealed AI’s Limitations.
Many executives already use gen AI as a thought-partner and c0-strategist. But are these tools reliable across markets? New research found that large-language models can mislead investors when operating outside their home information environments. In testing nearly 5,000 Chinese stocks, U.S.-based ChatGPT issued systematically more bullish forecasts—and made larger errors—than China-based DeepSeek, largely due to that system’s lack of diverse Chinese news coverage. The findings underscore a strategic imperative: treat AI as a powerful but partial lens, rigorously supplementing its outputs with local data, domain expertise, and cross-model validation.
Many executives already use gen AI as a thought-partner and c0-strategist. But are these tools reliable across markets? New research found that large-language models can mislead investors when operating outside their home information environments. In testing nearly 5,000 Chinese stocks, U.S.-based ChatGPT issued systematically more bullish forecasts—and made larger errors—than China-based DeepSeek, largely due to that system’s lack of diverse Chinese news coverage. The findings underscore a strategic imperative: treat AI as a powerful but partial lens, rigorously supplementing its outputs with local data, domain expertise, and cross-model validation.
What's Your Reaction?
:max_bytes(150000):strip_icc()/GettyImages-2187410260-5afe56694791404f99c253dd8d5abe6a.jpg?#)
