Formal Model of Trustworthy Artificial Intelligence Based on Standardization
Eduard Manziuk, Olexander Barmak, Iurii V. Krak, Olexander V. Mazurets, ТЕТЯНА СКРИПНИК · 2021
The widespread and rapid distribution and application of artificial intelligence (AI) systems requires the development of formalized approaches and the construction of basic principles for the functioning of domain areas of AI use. This need is embodied in the development of recommendations and standards to obtain maximum benefits from the use of AI and minimize possible risks. The regulatory framework is being built on a human-centric basis. Accordingly, the developed standards should form the basis for further activities aimed at the use of AI and be applicable at all stages of creating practical solutions. Therefore, an important stage is the formalization of requirements, principles and provisions of legal and ethical norms in the form of practical template approaches for practical application. With this method, models and ontology of standardized concept of AI credibility are developed within the research. This made it possible to identify the main concepts that allow forming a position of trust, are a meaningful part of the concept of trustworthy AI, determine the need for its existence and pose a threat to it. On the basis of ontology of the domain area, models were developed and further decomposition of structural substantive concepts was carried out. In the future, the characteristics of the concept of trustworthiness formation are defined.