click click click work work work by Linus Ziegler (
@_zglr ) and Fionn Heron (
@heshhound )
Student project from the course Perspectives in Design
While Artificial Intelligence is often portrayed as an autonomous and self-sufficient technology, most AI systems are deeply reliant on manual training and annotation work, essentially creating an illusion of machine intelligence by mechanizing human cognition and crowdsourcing workforce from the global south. In that way, the development of Artificial Intelligence is driven by a neo-colonial infrastructure of knowledge extraction, sustained by the exploitation of remote, dispersed and poorly-paid “click workers.”
In an attempt to shed light on click work as a foundational yet largely invisible component of Artificial Intelligence, click click click work work work explores how this already exploitative form of digital labor is further intensified through the use of gamification tactics.
Players are asked to mask objects within images and are rewarded based on speed and accuracy, recreating the conditions of real-world annotation tasks. The images used in the game are sourced from publicly available AI training datasets such as LAION-5B and COCO, highlighting the often ambiguous origins of the visual material used to train machine learning systems.
By framing click work as a “game you cannot win,” the project exposes how exploitative labor structures can be obscured through playful interfaces and productivity mechanics. In doing so, it invites players to reflect on the hidden human infrastructure behind AI systems and the labor-intensive processes that underpin seemingly intelligent technologies.
