The article is devoted to the conceptualization of the artificial intelligence phenomenon in the financial domain. It highlights the challenge posed by the absence of a cohesive theoretical framework and precise definitions for the concept of AI finance. This deficiency impedes a systematic examination of how these technologies influence the financial system. The purpose of the study is to develop a conceptual approach through the systematization of existing perspectives, the formulation of a scientific definition, and the creation of a formal model for measuring AI finance. As a result, three paradigms of understanding AI finance are identified: instrumental, transformational, and agent-based, each varying in the level of system autonomy. The study proposes defining AI finance as a system of financial relations with varying degrees of decision-making autonomy. To quantitatively evaluate the impact of AI on financial processes, a model named AI Labor Efficiency Impact (AI-LEI) is introduced, which considers the variations among the paradigms. This model facilitates the measurement not only of economic efficiency but also of the extent of structural changes occurring within the financial system.