The Case for Usable AI: What Industry Professionals Make of Academic AI in Video Games

Pfau, J; Smeddinck, JD; Malaka, R

doi:10.1145/3383668.3419905

The Case for Usable AI: What Industry Professionals Make of Academic AI in Video Games

Lookup NU author(s): Dr Jan Smeddinck ORCiD

Downloads

Accepted version [.pdf]

Licence

This is the authors' accepted manuscript of a conference proceedings (inc. abstract) that has been published in its final definitive form by Association for Computing Machinery, 2020.

For re-use rights please refer to the publisher's terms and conditions.

Abstract

Artificial intelligence (AI) is a frequently used term - and has seen decades of use - in the video games industry. Yet, while academic AI research recently produced notable advances both in different methods and in real-world applications, the use of modern AI techniques, such as deep learning remains curiously sparse in commercial video games. Related work has shown that there is a notable separation between AI in games and academic AI, down to the level of the definitions of what AI is and means. To address the practical barriers that sustain this gap, we conducted a series of interviews with industry professionals. The outcomes underline requirements that are often overlooked: While academic (games) AI research tends to focus on problem-solving capacity, industry professionals highlight the importance of "usability aspects of AI'': the ability to produce plausible outputs (effectiveness), computational performance (efficiency) and ease of implementation (ease of use).