The use of artificial intelligence (AI) in the finance world is taking another leap forward, as the chatbot ChatGPT is able to reliably determine whether stock moves are prompted by news headlines. This first wave of research applies AI to the world of finance and the results have already shown that the technology is proving its worth.
Two research papers have been published this month that tested ChatGPT on finance-related tasks – one in judging whether Federal Reserve statements are dovish or hawkish, and the other in deciding whether news headlines are good or bad for stocks. The chatbot demonstrated excellence in both, indicating a huge development in AI that can help turn written messages such as earnings transcripts and tweets into trading signals.
Man AHL, a company engaging in machine learning, is already using the natural language processing that the chatbot utilizes to detect patterns in texts. Head of machine learning at Man AHL, Slavi Marinov, approves of this technology, stating that the hype is a reality. The first paper, entitled “Can ChatGPT Decipher Fedspeak?”, conducted by Anne Lundgaard Hansen and Sophia Kazinnik of the Federal Reserve Bank of Richmond, confirms the chatbot’s accuracy in comparison to Google’s BERT and manual categorizations done through dictionaries.
OpenAI, which developed the technology behind the chatbot, recognized the potential of AI in the financial sector, hoping to provide the cutting-edge natural language processing that ChatGPT offers. With such positive and promising results from the early stages of research, it appears as if OpenAI’s ambitions are coming to fruition.
