New And Surprising Ways to Be Mean

By C. Guckelsberger, Christoph Salge, Julian Togelius. 2018

16 citations recorded by the source corpus.

Creating Non-Player Characters (NPCs) that can react robustly to unforeseen player behaviour or novel game content is difficult and time-consuming. This hinders the design of believable characters, and the inclusion of NPCs in games that rely heavily on procedural content generation. We have previously addressed this challenge by means of empowerment, a model of intrinsic motivation, and demonstrated how a coupled empowerment maximisation (CEM) policy can yield generic, companion-like behaviour. In this paper, we extend the CEM framework with a minimisation policy to give rise to adversarial behaviour. We conduct a qualitative, exploratory study in a dungeon-crawler game, demonstrating that CEM can exploit the affordances of different content facets in adaptive adversarial behaviour without modifications to the policy. Changes to the level design, underlying mechanics and our character’s actions do not threaten our NPC’s robustness, but yield new and surprising ways to be mean.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Sociology and Political Science

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Shimon Whiteson
2. Sergey Levine
3. Tabish Rashid
4. Yoshua Bengio
5. Pieter Abbeel
6. Mohit Bansal
7. Anuj Mahajan
8. Kyunghyun Cho

In Computer Vision and Pattern Recognition

1. Nando de Freitas
2. Caglar Gulcehre
3. Shimon Whiteson
4. Angeliki Lazaridou
5. Tabish Rashid
6. Pedro A. Ortega
7. Joel Z. Leibo
8. Prasad Tadepalli

In Sociology and Political Science

1. Pedro A. Ortega
2. Joel Z. Leibo
3. Shimon Whiteson
4. Angeliki Lazaridou
5. Tabish Rashid
6. Edward Hughes

Citations in the corpus

• MAVEN: Multi-Agent Variational Exploration (2019, background): “[15] maximise the empowerment between one agents actions and the others future state in a competitive setting.”
• Optimal Policies Tend To Seek Power (2019, background): “In a two-player agent / human game, minimizing the human’s information-theoretic empowerment [Salge et al., 2014] produces adversarial agent behavior [Guckelsberger et al., 2018].”
• Social Influence as Intrinsic Motivation for Multi-Agent Deep Reinforcement Learning (2018, background): “Guckelsberger et al. (2018), in pursuit of the ultimate video game adversary, develop an agent that maximizes its empowerment, minimizes the player’s empowerment, and maximizes its empowerment over the player’s next state.”