A new study from the University of Chicago shows thatLarge Language Models GPT-4 Able to outperform humans in predicting future earnings growth of companiesAnalyst, and the AI model only used the company's financial statements without any additional information.
Image source: Pexels
Financial analysts have traditionally relied on their expertise and experience to assess a company’s financial health and predict future earnings, but this study shows that AI models can do the task just as well, or even better.
It is understood that the researchers provided the model with anonymized financial data, including balance sheets and income statements, and asked it to predict future profit growth. Even without any additional information,GPT-4 was also able to achieve an accuracy of 60%, while the typical accuracy range for human analysts was only 53% to 57%.
Part of GPT-4’s success is due to its use of “thought chain” prompts, which guide the AI model through the analysis process, helping it identify trends, calculate ratios and synthesize information to make predictions.
Researchers believe that large language models have significant advantages over human analysts due to their vast knowledge base and strong pattern recognition capabilities.
Although some experts have cautioned that the benchmark used in this study may not be the most advanced financial analysis model, the overall findings are still encouraging. The ability of general AI models to outperform specialized financial models and even human experts suggests that large language models have the potential to transform the field of financial analysis.