Project Description

The continuous advancement of artificial intelligence (AI) is opening up new possibilities for having even complex decisions handled with computer assistance or entirely by a computer. From a scientific perspective, this increasingly raises the question of which types of decisions in production management exist where an AI system is on par with—or perhaps even superior to—humans. When applied to other fields, this question has already been answered in a similar way for games such as chess, Go, or Jeopardy.

The goal of this research project is, on the one hand, to identify those production management decisions that can be handled by AI-based approaches using reinforcement learning (RL), and, second, to determine whether the human-machine duel is a suitable method for evaluating the performance of AI-based approaches in production management.

Particularly in the case of RL, advances in AI research show that even artificial agents can match or even surpass the performance of human players in complex scenarios. The ManExMa research project therefore takes the approach of applying existing insights from the human-versus-“machine” (AI instance) duel to production management. To this end, various production management problems and decision-making scenarios will be modeled in a simulation environment, which can then be tested in a game-like setting. Finally, the performance of humans and AI will be evaluated and compared. To ensure that the results obtained from the simulations are applicable in a real-world setting, they will ultimately be validated in a real production environment.

On behalf of the IPEM Chair at the University of Siegen, our research associate Luisa Stracke
is serving as the operational project manager for this research project.