Artificial Intelligence vs Human: Which is Better in Finance?

When technical analysis is mentioned, the first thing that comes to mind is the world of finance, and today, everyone who enters the world of finance, especially the stock market, begins to research how to analyze data. Yes, there are famous financial analysts who are very popular and what they say is often true. So how is artificial intelligence doing in this regard?

Researchers at the University of Chicago say that large language models, and especially GPT-4, can analyze financial statements with higher accuracy than humans.

Financial Analyst Success from Artificial Intelligence!

The researchers first fed GPT-4 with anonymized and standardized corporate financial statements. Two methods were then used to test the LLM’s capabilities. The first were simple requests that led the chatbot to analyze expressions and predict future earnings.

The second was to use a “chain of thought” (CoT) prompt that taught the AI ​​model to imitate financial analysts. With CoT, GPT-4 was asked to identify notable trends, calculate key financial ratios, and establish expectations for future earnings.

The first method, simple prompts, did not produce significant results, but CoT achieved an accuracy rate of 60.30 percent, higher than the average human analyst.

As we mentioned at the beginning, in order for artificial intelligence to produce good results, it is necessary to constantly feed it with new information. Therefore, the common view of researchers is that instead of saying “artificial intelligence is better or humans are better”, the two complement each other.

While the LLM excels in areas where humans may be inefficient or biased, humans add great value to analyzes where additional context is needed. The recently introduced GPT-4o has already managed to impress many people with its capabilities.