Understanding People Opinion on Artificial Intelligence Ethics through Machine Learning-based Sentiment Analysis

Abstract

Artificial Intelligence (AI) ethics is so near to human existence. It contains a collection of beliefs, concepts, and methods that utilize generally recognized moral standards to govern moral behavior in developing and using AI technology. Besides providing many advantages to people, the development and use of AI also pose risks that may harm the humankind. At the crossroads between the need for AI and the risks, we want to know what people think about AI. For that purpose, we devised a sentiment analysis system powered by Naive Bayes Classifier and Term Frequency-Inverse Document Frequency (TF-IDF) methods. After analyzing 1.138 data taken from YouTube and Twitter which it is categorized into three opinion labels, namely positive, neutral, and negative, the system can achieve accuracy of 71% with average precision of 66%, average recall of 71%, and average F1-Score of 64% through K-Fold Cross-Validation.