Data Ethics: Issues related to data biases and the application of traditional ethical theories on AI Ethics

The purpose of this article is to present and evaluate some issues related to data ethics in the era of artificial intelligence (AI). First, a philosophical definition of AI is attempted following the classic portrayal of AI presented by John Searle. This sets the frame for understanding AI and its basic components, machine learning, and big data. A distinction of the basic, traditional ethical theories (deontological ethics, utilitarianism, and virtue ethics) follows to draw an ethical schema via which to assess the discussion of data ethics in this paper. Then, the basic biases that are relevant to data gathering and processing are thoroughly presented with some significant examples to enhance understanding of these issues. After explaining the biases related to data, an assessment of how these could be eliminated follows based on the simple ethical schema presented at the initial stages of the article. None of these ethical theories suffice to eliminate data biases. However, their importance lies in the fact that they can be used as ethical methodological tools to assess data bias and understand the ways in which it may hinder the presumed objectivity of AI.