Understanding the Research
Large language models (LLMs) like ChatGPT and Google Gemini are being explored for their potential to assist individual investors in navigating complex market data. Researchers Yi Cao and Long Chen, along with their colleagues, have developed a study that investigates how LLMs can identify “peer firms,” or competitors within an industry. Their findings suggest that these AI tools can streamline the process of data aggregation and analysis, making it easier for investors to evaluate firm values based on comparable companies, much like real-estate appraisals.
Key Findings
- The study utilized Google’s Bard (now known as Gemini) for its extensive pre-training data and parameters.
- The researchers focused on large, publicly listed companies, analyzing over 300,000 focal firm-years.
- On average, the LLM identified around seven peer firms per focal firm, aligning with SEC guidelines.
- A comparison showed that the LLM’s peer lists overlapped over 40% with those generated by human experts and outperformed traditional identification systems like SIC codes.
The Bigger Picture
The implications of this research are significant for individual investors, who can benefit from the efficiency and accessibility that LLMs provide. While LLMs demonstrate a strong capability in data processing, the study also highlights that they are not infallible. Investors must weigh the pros and cons of using AI tools, as they cannot be held accountable for the information they generate. As the technology evolves, understanding its limitations will be crucial for making informed investment decisions.
