In this work we present the construction, empirical validation and experimental application of a web based system for teaching topics of Business Administration. The same concepts can be easily extended to other formative areas. The system realizes a cooperative behavior of human agents (learners) who interactively take decisions for a simulated profit oriented enterprise.
The technical design is based on System Dynamics and Artificial Agent modeling (Scholl, 2001).
An agent based framework is applied to the model in the form of virtual tutoring system for the learners; the intelligent agent learns through a trial and error technique, based on Reinforcement Learning paradigms, by practicing the system. After the trial period, it computes an internal model of how the simulation works (in particular of the cause/effect relations among the decisions taken and the observed results) and then it’s used as a decision support system for the human learners, during the game play.

In the simulated model, the users take a number of core decisions at the beginning of each month; the system, based on Forrester’s System Dynamics, generates a set of reports, typical of Management and Enterprise analysis. The users, by reading and analyzing them, have to track down the influence of the single decision – or even better the aggregate effects coming from two or more decisions – on the synthetic results, representing the monthly performance of the whole enterprise.
The main system is split into subsystems, which metaphorically represent enterprise areas, such as acquisition of resources, production of goods and services, capital investments, research and development, marketing and so on. External stimuli are to be intended as responses of markets and supply. According to the traditional use of the above quoted modeling, the design of a whatever economic system being simulated provides a discussion of its results “at the end” of simulation, and the traditional transfer of knowledge suffer from the paradigm of “coeteris paribus” i.e. teaching is concerned with the behavior of just given variables at a time, while keeping the remaining ones “still”. On the contrary, in the presented approach, the learner (or better, the team of learners) “drives” the whole system by his own decisions during the decision making process, learns through the web interaction and the process of “role playing” within the group, and then gets a set of reports, computed by the virtual agent.

This system has a nature of “behavioral game”, where learners are motivated by the attainment of goals like profit, enterprise development, and so on. In fact, this game offers a natural domain for empirical observation of the dialogue among human agents (Airenti, Bara and Colombetti, 1993) but at the same time uses artificial agents to monitor the system.
A working framework is described in detail, as long as some operative applications in schools and universities. Different agent based paradigms –namely reactive and cognitive ones - are then described as a future development for the system.