Building Ethical AI from News Articles

Abstract

Improving performance of artificial intelligence (AI) has been the most important focus of AI studies. As a result, AI is expected to replace what considered to be done by human intelligence. Additionally, human societies are willing to accept AI agents as members to collaborate with. For AI agents to be successfully integrated into human societies, the agents must understand important values in the societies. Ethics is a system of values that sustain the societies and it is considered as the most important. Therefore, there should be a way for AI agents to understand ethical values of human societies. In this paper, we show how to extract ethical values and moral values of the age from news articles using natural language processing. Our result shows that from newspaper articles ethical and moral values could be extracted and modeled for AI agents to refer. In conclusion AI can calculated ethicality of text by itself. In the future, we can develop more ethical autonomous AI agents.