人工智慧應用於人力資源之實證研究: 以AI九型人臉辨識為例

近年來深度學習技術的進步讓人工智慧廣泛的應用在協助人們處理生活及工作上的大小事，包含運用人臉辨識協助人力資源的差勤管理、甄選等工作。本研究以實驗法進行研究，旨在探討AI與真人在進行九型人格之性格測驗的判斷差異；研究方法分別以公立就業服務機構的就業服務員及民眾為研究對象，透過實驗設計由就業服務員與AI對民眾進行性格判斷，再將實驗結果以Probit模型進行實證研究，以討論其中差異並據以推論AI與真人之性格判斷成效差異並做成結論。本實驗共計回收170份有效樣本數，經實證研究結果發現：（一）人工智慧判斷性格結果正確性有待提升；（二）九型人格為第五型者（觀察者）判斷性格的正確性顯著高於其他類型者；（三）至就業中心之目的為「領取津貼」的受測者，其性格較容易被判斷正確。

關鍵字

人工智慧；人力資源；就業服務員；九型人格；人臉辨識

並列摘要

The advancement of deep learning technology has made Artificial Intelligence (AI) widely used to assist people in many aspects of life and work, including the use of face recognition to assist the recruiting task for human resources management. This study uses an experimental method to study the differences in the performance of personality tests between AI and real people. The research method is to take the Employment Counselors and the individuals coming for public employment service as the research objects. The first step of our experimental design is to have the Employment Counselors and AI to conduct personality judgments. This study then conducts empirical research on the experimental results based on the Probit model to discuss the differences in the performance of personality tests between AI and real people. Based on the empirical results, we can infer the differences in the personality judgment effectiveness between AI and real people and draw our conclusions. A total of 170 valid samples are used in this experiment. The conclusion from the empirical research are as the following: (1) The accuracy of artificial intelligence judgment of personality needs to be improved; (2) People whose Enneagram is the fifth type (The Wise Person) are significantly more accurate in judging the personality than other types; (3) Those who go to the employment center for the purpose of receiving subsidies are more likely to get correct personality judgement.