- A study found that ChatGPT reasons similarly to humans when deciphering Fed statements.
- GPT-4, released in mid-March, was more likely to agree with humans than GPT-3.
- A separate study showed that the chatbot can also make solid stock market predictions based on company news.
OpenAI’s ChatGPT has shown its ability to decipher the meaning of Federal Reserve statements and has also shown predictive abilities of the stock market in response to headlines, according to separate studies.
Richmond Fed researchers said in an article that ChatGPT can reason similarly to humans when decoding the Fed language.
Testing both ChatGPT-3 and GPT-4, the study took 500 sentences from Fed policy statements and asked the AI to sort them into five categories: dovish, mostly dovish, neutral, mostly hawkish, and warmonger.
Even without prior training, the AI outperformed BERT, another language model used in finance, and was closely aligned with how four human reviewers categorized sentences. In addition, she was able to provide a reasonable justification for her choices:
“The sentence indicates that the committee expects to begin implementing its balance sheet normalization program soon, which is a clear signal for monetary policy tightening, as long as the economy develops as expected,” the report wrote. GPT-4 about a statement deemed warmongering.
GPT-4, released in mid-March, was more likely to agree with humans than GPT-3.
In a second studyUniversity of Florida researchers made financial headlines after October 2021, when ChatGPT’s knowledge base ends, and prompted the AI to make investment recommendations based on it.
This way, ChatGPT could be tested on how the news would move a stock without having prior information about what actually happened. The results showed a strong correlation between ChatGPT responses and stock market movements.
For example, the authors asked ChatGPT to decide what the headline “Rimini Street fined $630,000 in case against Oracle” meant to Oracle.
He replied that investor confidence in Oracle would increase because the decision could improve the company’s ability to protect its intellectual property.
“These results confirm the predictive power of ChatGPT sentiment scores and underscore the potential benefits of integrating LLMs into investment decision-making processes,” the researchers concluded.