This paper shows a change in the paradigm of medium-oriented training of complex systems operators on simulators using high-precision models of real control systems and their application environments. Evolution of the technogenic environment leads to an increase in the complexity of professional activity, which hinders the implementation of simple algorithmic learning technologies. Forecasts to reduce the role of man in ergatic systems are not justified. New ergatic systems using artificial intelligence technologies require from operators qualities that exceed human capabilities, which requires an examination of projects and an assessment of user control complexity. The reasons for the non-classical “complex learning” methodology, which takes into account the student’s self-organization and orientation in a complex dynamic context, are shown. Promising methodological approaches to the development of simulators that implement symbiotic medium-oriented learning are considered.