Response: Intelligent agents, simulation, and gaming

October 11, 2012

Intelligent agents, simulation, and gaming by Levent Yilmaz, Tuncer Ă–ren, and Nasser-Ghasem Aghaee is a brief introduction to more focused papers regarding the design and implementation of various artificial agents for use in gaming and simulation environments. The authors focus on AI-based, or “agent-supported simulation” (developing generative behaviour of a model or system through AI techniques), and discusses ways in which these agents can be implemented into simulation systems in relatively abstract terms.

The concept of “agent simulation” is distinguished from “agent-supported simulation,” by which agent simulation involves the design of agents as a means of imitation (e.g. representing a human and their associated behaviours), while agent-supported simulation utilizes similar techniques to improve the overall system’s ability to identify and solve problems.

Agent-support simulation seems to be the more interesting of the two, as successful implementations of such technology can generate varying environments and scenarios based on the user’s participation and performance. While the paper discusses military simulation as one such example, where a system’s AI can generate alarms based on pattern recognition (for example, if a player is overly ambitious in their actions, the simulation might punish them by calling for more enemy fire), I foresee other uses of this technology in education. For instance, perhaps a computerized system could monitor a student’s performance on tests over the course of a term, and based on the student’s answering styles (and their successes or failures), the system might offer an alternative lesson plan or teaching style. If this type of ‘adaptive teaching’ could be applied in a gamified or simulated context (which is the article’s focus) as a supplementary mode of learning, we might see dramatic improvements in a student’s performance over time.

Overall, this paper acted more as a gateway to additional reading rather than a way of providing insight into practical uses of agent-support simulation. I’m interested in seeing how intelligent systems and simulations can be integrated with existing and aging infrastructures to enhance the ways we learn about our world, and in turn, how those changes will affect the intelligent system’s learning.