Source | AI Frontline Author | Tina
On May 24th, Ali Dharma Institute released a research paper on the cost accounting of replacing data analysts with GPT-4.
The paper states: “In addition to the comparable performance between all data analysts and GPT-4, we can notice that the time taken by GPT-4 is much shorter than that of human data analysts. We assume that there are about 21 8 hours per day, and calculate the cost per instance in terms of dollars based on the average time spent by each level of data analyst. GPT4 costs about 0.71% of the cost of a junior data analyst and a senior 0.45% of data analyst costs."
This new paper from Ali Dharma Academy and Nanyang Technological University in Singapore mainly discusses whether GPT-4 can do a good job as a data analyst. The title of the paper is "Is GPT-4 a Good Data Analyst?"
The paper focuses on the following capabilities of GPT-4 as a data analyst:
-
Generate SQL and Python code;
-
Execute code to obtain data and charts;
-
Analyze data and draw conclusions from data and external knowledge sources.
In Section 5.3 "Main Results", the paper states that the performance of GPT-4 is comparable to a human who has worked in the financial industry for 6 years on most indicators, the accuracy is lower than that of humans, but the complexity and consistency indicators higher than humans. In comparison with another analyst with 5 years of work experience, GPT-4 lost to humans in terms of correctness of information, aesthetics of charts, and complexity of insights. When compared to junior analysts with 2 years of work experience, GPT-4 performs better in correctness and can complete more work.
Although GPT-4 is currently less accurate than humans in advanced data analysis tasks, it is cheaper.
In other words, according to the experimental results and analysis, GPT-4 has comparable performance to humans in data analysis, but whether it can replace data analysts requires further research to draw conclusions.
Paper address: https://arxiv.org/abs/2305.15038