Abstract:

Artificial intelligence is rapidly being introduced into all spheres of life: economic, social, production and technological. University students enrolled in economics and management programs are now, in most cases, pursuing serious data science programs, including the use of machine learning methods to inform management decisions in economics and public administration.

The problem arises of forming a trajectory for training students in methods of working with data and their analytics, ensuring the synchronous formation of the skills and knowledge of future professional analysts with the speed of implementation and the capabilities of artificial intelligence, and more effectively - the formation “ahead of the curve.” The article proposes a conceptual approach to the formation of an “ideology” and trajectory for training specialists in data analytics to ensure their understanding of the solutions offered by AI. In the very near future, conflict between the “train of thought” of humans and AI will be a very common phenomenon. The article explains that an expert analyst and AI can and should understand and strengthen each other, and this requires a special learning path.

We see a solution to this issue by explaining to the student the paradigm of changing predefined analytics to unpredefined analytics as they move through the stages of ascendant analytics.

Non-predetermined (advanced) analytics methods based on machine learning algorithms arise at the stage of diagnostic analytics, increasingly providing all subsequent stages of ascendant analytics. In this case, the identified structures, sets of significant factors and patterns of their influence, and other results of data analytics are not based on hypotheses formed in advance by the researcher. They arise as a result of skilled work with data on the principles of data mining. As a result, the Data-Information-Knowledge pyramid ceases to be a peaked pyramid. The upper part (Knowledge) expands significantly relative to the researcher’s initial knowledge.

This approach to developing a learning path for analysts should be complemented by an explanation that for an analyst to be receptive to an AI solution, it is necessary to provide the AI with all possible information. The main thing is the willingness to interpret the proposed solution, understanding that it can be based both on “visible” (predetermined) hypotheses and on revealed latent connections, the logic of which may be unfamiliar and even strange. The question is whether to influence the AI to change this logic or accept it by finding an explanation for it.

Keywords:

Analytics, cognitive computing, artificial intelligence, non-predefined approach.