Inspirations
Ian Chang
Ian Cheng’s work with machine learning is an important inspiration to us through its exploration of the ability for independent and “free” digital agents to produce meaningful narratives through their interactions with their environment and one another. Cheng creates agents with unique behavioural traits through the process of machine learning and then thrusts them into varied environments populated with other unique agents, each with their own motivations and prerogatives. He then captures the stories that develop over the course of their lives, shedding light on the narrative power of uncontrolled, pseudo-sentient AI.. This source was found early in our development, and changed in some ways the way we perceive AI..
Multi-Agent Hide and Seek by OpenAI
This source is crucial to our projects development and has been deeply informative both on technical and theoretical levels. In Multi-Agent Hide and Seek, Open AI has created a seminal work that explores, for the first time in an open source project, emergent behaviour in ML agents taught using agent versus agent reinforcement learning. We have relied heavily on the pioneering insights of the Open AI team especially via their research paper, which contains a variety of development graphs breaking down their approach and stages of development. Though the technicalities of our projects are on completely different frameworks, the logic and theoretical approaches are similar enough that we can glean useful information.
“Infranet” Haru Ji
This installation by Haru Ji is an important inspiration to us in that it elicits a sense of omniscience in the viewer by way of the fact that they can see the entirety of a vast dynamic world in the format from a digital, somewhat dioramic perspective. This sense of an all-seeing presence in the form of the viewer harkens to our theme of surveillance, informing our creative perspective on the concept and furthering our understanding of the subject as it relates to a work of a similar technical nature. Furthermore, “Infranet” is a vast ecosystem of ML entities who interact and compete with each other, a dynamic that speaks to our vision for this project. The piece has in some ways served as a catalyst in the development of our desired final aesthetic as well, as a piece with a simplistic and dark design language, we feel that it communicates its intent without distracting the viewer with loud visual components. We are inspired by “Infranet's” ambitious technical scope, especially in the field of the use of neural networks for AI training, and it made us more confident as a team during the early stages of the development of the AI component.
-
Share