Abstract

The success of machine learning algorithms generally depends on data representation, and we hypothesize that this is because different representations can entangle and hide more or less the different explanatory factors of variation behind the data. Although specific domain knowledge can be used to help design representations, learning with generic priors can also be used, and the quest for AI is motivating the design of more powerful representation-learning algorithms implementing such priors. In this review paper, statistics of AI incidents and areas are presented along with the social impact. Using the online AI Incident Database, some areas of AI applications have been identified, which shows unethical use of AI. Applications like Language and Computer vision models, intelligent robots and autonomous driving are in top ranking. Ethical issues also appear in various forms like incorrect use of technology, racism, non-safety and malicious algorithms with bi-asness. Data collection has helped to identify the AI ethical issues based on Time, Geographic Locations, Application Areas, and Classifications.