An Experiment to Find the Cause of Information System Errors Using Image Captioning AI

Abstract

Many companies in Japan promote DX. This clinic provides medical examination services. In this clinic, routine tasks are automated using RPA. However, the information systems stop due to various errors. Since this clinic does not have a full-time information system administrator, the clinic staff must resolve these errors themselves. The increased time required for staff to resolve errors is problematic. Our goal is to enable the clinic staff to resolve the causes of errors using image captioning AI without technical knowledge of information systems. As a first step, we experimented with AI to find the causes of two types of errors in the information systems using RPA. The results showed that the output sentences sometimes correctly explained the causes of the errors, but at other times provided incorrect explanations.