Topology Unknown (2025); Single Chanel Video

Topology Unknown is an AI-driven experimental game prototype in which the player takes on the role of an explorer on the AI’s black-box expedition to traverse and map the uncanny, unsolvable multidimensional landscape. Unlike traditional first-person view games and existing generative game map design, where the map is persistent and constant once generated, Topology Unknown’s world is unstable and abstract with the player's controls directly influencing how the AI generates the landscape in real time. This pioneering solution is achieved by mapping players’ movements to AI's latent-space walk technique (Schaerf, 2024), driven by AI’s emergent properties. Therefore, the AI landscape will never be the same when viewed from different angles, at different moments, or during different play sessions: no two players will experience the same visuals and landscape. We observed that distributed agency, having moved from the designer to a triad designer/player/AI, produces conversational play with further possibilities of researching on human-machine connections through playful acts (Kac, 2004). In this context, embracing uncanny affordances as the introduction to the interpretative act of communication, rather than defects, expands possible pedagogical approaches to game aesthetics.