HUMAN-MACHINE MEMORY
This catalogue of making consists of a series of experiments that aims to differentiate between human and machine memory. Using different materials and tools, I investigate the relationship between human memories, represented as images and texts, and machine memories, represented as processable data.
These experiments are then curated and compiled into a website alongside a narrative that supports the outcomes.
PROMPTING WITH LLAVA
A prompt activity using AI model to interpret human memories collected in the form of images and text. Users were asked to draw their childhood memories and describe them in one to two sentences. These drawings and descriptions were then fed to LLAVA for image analysis and MidJourney for image generation.
QUANTIFYING IMAGES
An exploration of image encoding using RGB values and ASCII characters. The experiment extracts an image's RGB values, manipulates them with ASCII text, and feeds the result into ChatGPT to see how it reconstructs the image based on its "understanding."
SEGMENTING IMAGES
This experiment involves segmenting images based on RGB color values. The segmented images are then fed into ChatGPT for image analysis to reflect on how objects in image are interpreted as groups of colored pixels.
QUANTIFYING MOVEMENTS
This experiment uses an accelerometer to convert movement data into RGB values, creating visual representations of images. Meaningful movements, like a hug, are recorded to explore how machines could perceive these actions through numerical color values.
VISUALIZING LATENT SPACE
This experiment aims to visualize the latent space interpolation concept in machine learning. By using texts and images as an analogy to understand the abstract idea of latent space, I created a website and video to illustrate the concept based on my narratives.