A Flexible Multi-Agent Systems Task Environment for Simulating Hybrid Intelligence

Hybrid Intelligence (HI) attempts to address and solve cognition-intensive tasks by combining Artificial Intelligence (AI) with human intelligence. Due to the novelty of the HI concept, there are few proven patterns for the optimal design of a Hybrid Intelligent System (HIS). To facilitate HIS simulations, we propose a framework to describe and simulate complex tasks in a multi-agent system. The tasks are solved collaboratively by teams of agents, simulating characteristics of actors with human or artificial intelligence. The prototype of a specification-driven task environment was evaluated using two sample applications. The results confirmed the strengths of the concept with regard to the study of different team constellations and cooperation patterns. Extensions would allow modelling of scenarios with higher complexity and more realistic simulation of human behaviors.