인공지능 연구 동향 및 신뢰성 분석: 자율주행차와 신뢰할 수 있는 인공지능 개발안내서를 중심으로

This study explains AI trustworthiness standards in developed countries and Korea were analyzed, and the 'Trustworthy AI Development Guide', which most specifically expresses trustworthiness requirements, was selected as the trustworthiness standard. And, based on the selected criteria, 40 papers on artificial intelligence of autonomous cars were analyzed, and the contents of individual studies and trends in four fields (learning data, model and algorithm, system, and interface) were identified. Among the four fields, in the researched papers of artificial intelligence learning data, a large number of contents that met the artificial intelligence trustworthiness requirements were found, and ethical requirements were explored through meta-analysis using SPSS. In the remaining three fields, there was very little or no content related to trustworthiness, and it was confirmed that the composition or contents of the paper as a whole were somewhat far from the trustworthiness requirements of the 'Trustworthy Artificial Intelligence Development Guide'. To verify the trustworthiness of artificial intelligence, it was proposed to describe the trustworthiness-related contents in the thesis, to develop detailed trustworthiness standards for each field, to establish a consulting and verification organization for trustworthiness verification, and to establish internationally acceptable trustworthiness standards.