Interfacing AlphaGo: Embodied play, object agency, and algorithmic drama

Area of Science:

• Science and Technology Studies (STS)
• Artificial Intelligence (AI)
• Human-Computer Interaction (HCI)

Background:

• Professional Go playing was long considered a benchmark for AI capabilities.
• AlphaGo's victory over a top human professional in 2016 marked a significant AI milestone.
• The event generated substantial media attention but limited critical analysis within STS.

Purpose of the Study:

• To analyze the 'AlphaGo show' as a technology demonstration, focusing on its social and technical aspects.
• To examine the significance of AlphaGo's 'move 37' and expert commentary.
• To explore the concept of 'scenic intelligibility' in the context of human-machine interfacing.

Main Methods:

• Reflexive video analysis of the AlphaGo exhibition match.
• Examination of expert commentary and media reporting.
• Analysis of embodied interaction and 'human/machine interfacing'.

Main Results:

• AlphaGo's 'move 37' and its presentation were contingent on intricate human-machine interfacing.
• Media portrayals of 'superhuman intelligence' relied on and obscured this interfacing work.
• 'Object agency' and 'algorithmic drama' emerged from skillful embodied play as a key interfacing practice.

Conclusions:

• The AlphaGo match demonstrated that AI achievements are deeply intertwined with human-machine collaboration.
• Understanding the 'AlphaGo show' requires scrutinizing the embodied practices of interfacing.
• The study informs how AI's perceived agency and dramatic narratives are constructed through such practices.