hana yoo





ANIMAL.MACHINE.PSYCHOSIS

    Chambers
    Hysteric C
    Splendour in the grass

LIMINAL ECOLOGIES

    Anthropology of Dead Body
    Rapture

news
CTM/transmediale Vorspiel 2021
Online Screening+Artist Talk | Berlin, Germany | 04.02. 2021

KI Camp_Futurium(online)
Virtual Exhibition | Berlin, Germany | 26.03 - 27.04. 2021

Heimat, Herz, Orte_Galerie Bernau
Group Exhibition | Berlin, Germany | 09.04 - 29.05. 2021


Mark
Chambers (work in progress)
Installation, dimension variable, 2021

Chambers, generated machine learning scene, 3D in Unity ML, non-stop playing, 2021
Ward No. 6, short film, HD, stereo, color, appx. Length 15mins, 2021




Chambers(work in progress), still, 2021 - The game agents(rats) are failing(escaping) from the training zone.




In 1948, a psychologist B.F. Skinner devised an operant conditioning chamber (known as a 'Skinner box’) to study animal behavior. Using a lever system that a rat presses to gain food inside a chamber, Skinner discovered that the food acts as a 'reinforcement' which induces rat's repetitive behavior. The premise that responses through conditioning shape actions became a pivotal model of Behavioral psychology, and Behaviorists consequently conducted various animal experiments to examine deeper on the learning process and problem-solving of the organism.




Operant Conditioning Chamber (Skinner’s Box), image source: http://www.appstate.edu



Now we can observe a rather complex model of learning with a Machine Learning (ML) system, which takes data and algorithm to make a machine ‘learn’ cognitively and utilize this learning in decision making. In spite of its complexity, the methods of ML vividly resonance with Skinner’s basic principle - such as Reinforcement Learning (RL) - which concerns how the agent takes action in an interactive environment in order to maximize reward. We are training machines and animals in an analogous learning model, whereas, rather obviously, its purpose and circumstance are divergent. Having concerned the outcome of training deeply intertwined with human behavior and the economics of ecology, we could raise the following questions - what sort of reduction has conducted by restricting certain environments in experiments with animals and machines? Are the anthropomorphized expressions to nonhumans such as ‘machine and animal learn’, ‘machine intelligence’ the part of ‘becoming’? Is there a new type of empathy through the technical apparatus? What do we learn from animals eventually, and from the AI system that we built?




        

Ward No.6, video sketch, 2021 




Following scientific-philosophical experiments from past to future, the project ‘Chambers’, an installation and a short film, investigates the possible thread of behavioral animal experiments to the current ML method - Reinforcement Learning. Having a laboratory rat and machine learning system as the anthropomorphized protagonists, the two characters in the short film create a meta-log about the ecology of empathy, invisible labor, algorithmic bias, and the anthropomorphism itself. Together with the film, the installation consists of a generative ML model, visualized by a game engine.











<Research map>




* This project is partially supported by German Federal Ministry of Education and Research and Society for Computer Science(Gesellschaft für Informatik e.V.) as a part of program KI CAMP