The Making of the 'AI World Cup': Attuning to the Ideal Human-Machine Configuration

• In Event: Machines and Humans: Risk, Fear and Collaboration

Abstract

This presentation will examine the making of the 'AI World Cup' as a new platform for human-machine collaboration as well as the attunements to concord the reality to this ideal. On December 2, 2017, the world's first AI World Cup was held at the Korea Advanced Institute of Science and Technology (KAIST) in South Korea. The AI World Cup was novel in a way that it took place inside the computer server, with the players and stadium all a part of a computer simulation program. The human participants were to submit the code that instructs the robots how to move and train themselves. Unlike previous events, such as human versus Deep Blue or AlphaGo which produced popular discourse on AI and robots as the 'confrontation' between human and machine, the organizer's idealistic vision of the AI World Cup was all about 'collaboration', which was inscribed in the design of the event. Within this vision, humans and seemingly "autonomous" machines were expected to work together to become an effective team, an effective human-machine assemblage.

However, as in other AI or robot events, the organizers and participants encountered disparities between the imagination and reality in making the AI World Cup. Well-organized collaboration between human and machine, in which humans and machines remain in their own designated places and are devoted to their own roles, was not achieved in the actual game. To hide the friction behind the curtain, attunements in various dimensions of the event took place, from technical issues regarding the compatibility of computer software and configurations to more fundamental matters of Deep Learning in AI technology. These attunements resulted as the human-machine collaboration considerably attuned to its idealized form. The place where humans stand within this relationship is being rearranged by repetitive attunements and endeavors to reconfigure the agencies of humans and machines.

Author

• Heewon Kim, Korea Advanced Institute of Science and Technology (KAIST)