Population Infinite Final Project: Artificial Superego

Concept

Transhumanism is a philosophical movement which advocates for the transformation of human capabilities, both physical and mental, through technology. The movement is becoming more relevant everyday with technology rapidly developing and becoming more involved in all aspects of our everyday lives and influencing our very identity as individuals and as a species. With recent advances in brain machine interface made by Neuralink with the vision of having a computer chip serving as a tertiary system for our brain on top of our existing cortex and limbic systems, I was inspired to contemplate the possibility of expressing subjective experiences using artificial technology.

My last exploration of machine learning image generation with styleGAN on RunwayML generated a group of pictures which are abstract yet resemble elements of the original photos, resulting in a dreamy aesthetic, as if the photos looked like they were generated subconsciously in one’s head. For the next project, I want to create an interface where the audience can input their own subjective experience to be converted into a machine learning generated photo which represents their experience.

Freudian psychoanalysis categorizes our brain into id, ego, and superego, with the superego being the mediator between one’s instincts and logical perceptions. Our project aims to generate a metaphoric representation of a combination of one’s subconscious input, and an artificial process by machine learning algorithm, thus the name “Artificial Superego”

The Test

To evaluate the user’s subjective experience, the interface will probe from three aspects meant to represent a portion of a range of possible subjective experiences: emphasy, self-awareness, and creativity. The three attributes will be tested using three questions. Based on the user’s response, an artificially generated image meant to represent the user’s subjective experience will display.

The Dataset

For the dataset, images of different colors and shapes are combined to represent different aspects of one’s subjective experience. Ideally we will come up with an algorithm to organize the generated images into different categories indicating the different subjective elements present in each picture, but for the first iteration, the project will include hand picked photos from the images generated on runwayML, manually categorized into different groups to be displayed based on the user’s test results.

Photos from the dataset are collected from image datasets found online, which include image libraries of various objects and environments such as trees, tools, mountains, flowers, and abstract paintings among others.

The Test

The test used to determine the user’s various characteristics are inspired by the Voight Kampff test from the film “Blade Runner” https://www.allthetests.com/quiz30/quiz/1386100782/Voight-Kampff-test-questions

and the Generalized Occupational Aptitude Test (G.O.A.T) from the video game “Fallout 3” https://fallout.neoseeker.com/wiki/G.O.A.T

There are three tests on the interface which are written to vaguely “test” the user from three aspects in order to direct them to the relevant image, the three aspects are self-awareness, sympathy, and creativity.

Photo standards

The photos generated from RunwayML were manually organized into different folders for each test results based on the following criteria:

Photos with geometric shapes and/or are black n white are associated with higher self awareness traits.

Photos which are soft & blurry looking, or are generally comfortable to look at are associated with higher sympathy traits.

Photos which are colorful and vibrant in the composition are associated with higher creative traits.

Team

I collaborated with Rui again for this project. Task distribution between us are as follows:

Work together: Conceptualizing, planning

Jack: Finding datasets, training and organizing machine learning generated pictures. Coming up with test questions

Rui: Coding the mechanics and designing the visuals of the interface. Taking the questions and photos from Jack and put into the interface.