préparation d'un numéro spécial de la revue Angles sur "digital subjectivities" 2017-2018
The last decade has witnessed a growing interest in the application of neural networks to the artificial intelligence systems. Drawing on the logic of biological neural network, artificial neural network (ANN) refers to a mode of statistical learning:
"Adaptive neural nets approach this, since they are capable of generating results that are not precoded. They automatically produce unforeseen results using feedback mechanisms to create resonance and interference between routines. In other words, what is coded is recursivity-machined self-referentiality. The digital processing becomes self-modulating: the running of the code induces qualitative transformation in its own loopy operation. Evolutionary digitality. Machinic invention.” (Massumi, 2002, p. 142)
With the growing popularity of online games and the emphasis on realist simulation, the AI systems become paramount to emulate a human-like behavior at the other side of the Internet connection. Neural networks have been one of the techniques to implement human like behavior in video games (Rabin, 2003). The implementation of neural networks in video games is long, requires lots of programming and is frowned with difficulties (choosing the right input, retraining the network in case the logic of the game changed, difficulties of debugging in case of problems). Given the statistical learning model on which ANNs are based, their training requires a large amount of human behavioral or physical data, “to integrate analog process into digital processing” (Massumi, 2002, p. 142). Their development is linked to the development of interconnected modular data collections made possible by the computerization of daily life (Greenfield, 2006) on the one hand, and to the development of big data techniques used to track players’ behavior in the virtual-actual world of video game (Bertero, 2014). Whitelaw equals the creativity of the human mind with the creativity of nature and matter, pointing at the question of the autonomous artwork and metacreation (Whitelaw, 2006, pp. 232–233). For Massumi the autonomy of affect and the concept of confound leads to the notion of art as cosmogenesis (Massumi, 2002, p. 173). Developing our earlier analysis concerning the concept of incomputable data in the context of video game creation (Charrieras, 2013), we will explore how the use of ANN in video games put into question the existence of games as rule-based system, or as authorship machines. Conceived as a digital-analog activity akin to world building, the emergence of some video games can be described as evolutionary.
About the author
Dr Damien Charrieras (assistant professor at the School of Creative Media - City University of Hong Kong) is a critical media theorist interested in tools and practices pertaining to media art, digital music, and video games. His papers were published in a large number of international academic journals, including Organized Sound(Cambridge), Screen and Human Relations (Sage). He is interested in the field of New Media Art, video games and electronic music. His work is situated at the crossing of new media theory, critical theory, sociology of art, cultural sociology and cultural industries. He uses the insights from new materialism and radical empiricism to critically assess the contemporary articulations of media technologies and art. He is especially interested in the question of self-perceptions, hybrid materialities, non-human agencies, affect and objects in relation to the digitally informed creative act. His current project investigates the diverse technologies used to conceive video games, especially game engines, from the perspective of media ecology and organology.
He is a member of Eastern Block, the Centre for New Media and Interdisciplinary Art (Montreal). He has been a visiting scholar at Brown University and worked as a researcher at the Institut National de l’Audiovisuel (Paris), at the Centre National de la Cinématographie (Paris) and at the Institut National de la Recherche Scientifique (Montréal). He is Stream Leader for the Media Cultures stream of the Master of Arts in Creative Media (MACM) at the School of Creative Media.
Neural nets; Game engine; Creative evolution; Radical empiricism.