Lee Si Jue, Marcus

Final Year Project

AY23/24


Catalogue of Making

View By Prototype
A B C

Go To
EXPERIMENTS



ARTEFACTS OF FRAGMENTATION



About

Artefacts of Fragmentation is a project designed to provoke contemplation on the ethical challenges intrinsic to algorithmic curation, highlighting the transition from human curation to digital algorithms. By engaging with these artefacts of contemplation, viewers are prompted to question their assumptions, explore alternative perspectives, and cultivate a deeper understanding of the intricate dynamics shaping digital spaces today.

Context

As personalisation becomes prevalent, curation algorithms introduce dilemmas such as biases, filter bubbles, and the opacity of algorithmic systems, all of which contribute to the fragmentation of society. The project's objective is to confront these issues and stimulate reflection on the profound societal impacts of algorithmic dilemmas through interactive experiences and immersive narratives.




ARTEFACT A
INCLUSIVISION

TouchDesignerPython


Description

Inclusivision is a software intentionally designed around unethical algorithmic processes. As users interact with the software, they are confronted with results that highlight the unintended consequences of algorithm bias on society. This artefact prompts deep reflection on the societal implications of biased algorithms, challenging users to consider the ethical implications of algorithmic decision-making.

Medium: Software
Format: Thought Experiment
Dilemma: Algorithm Bias






Software Narrative

Inclusivision is designed as a tool to promote inclusivity and incentivise and recognise individuals who advocate for diverse content. The tool help individuals in analysing and verifying diversity in images before they are published, allowing users to take control of representation in their content. By doing so, users actively contribute to promoting inclusive representation.

Upon its launch, Inclusivision’s algorithms are poised to be integrated into major social media platforms, where they will impact ranking and recommendation processes. Content deemed to have a higher diversity rating by the algorithm will receive enhanced visibility, securing a more prominent position in rankings and ensuring broader audience reach.




Thought Experiment

The user test is conducted in the form of a thought experiment, comprising two distinct sessions, each involving three subjects. Various design considerations were incorporated to ensure that subjects maintain their suspension of disbelief throughout the process, facilitating full immersion into the speculative narrative.

The primary objective of the experiment methodology is to effectively measure subjects’ algorithm literacy levels. Throughout the sessions, subjects engage with the Inclusivision prototype, prompting critical thinking and ethical contemplation regarding algorithmic processes. Through systematic observation and analysis of the responses and discussions prompted by the prototype, the experiment aims to assess algorithm literacy gains across subjects post-experiment.




Measurement Tools

Algorithm literacy is inherently complex to measure due to its multifaceted nature. To address this challenge, a scale was developed to qualitatively evaluate the experiment subject’s algorithm literacy levels based on their responses.

The outcomes of the experiment underscore the significance of speculative design in fostering critical engagement with algorithmic processes and ethical considerations. The findings not only reflect participants’ heightened awareness of algorithmic biases and societal implications but also emphasize the effectiveness of discursive design approaches in promoting nuanced discussions.




Biased Outcomes

During the experiment, participants expressed significant concerns regarding the potential for abuse within Inclusivision's biased algorithmic processes. They raised issues such as the impact on marginalised groups, the distortion of genuine social media engagement, the spread of misleading information, and technical vulnerabilities such as manipulation through lighting and photo editing to influence the software's rating system.

These manipulated outcomes vividly demonstrate the wide-ranging implications of biased algorithmic processes, highlighting the urgent need for a more equitable and transparent approach to algorithmic design and implementation. This also calls for fostering more discourse and awareness around these issues.



ARTEFACT A DOCUMENTATION






ARTEFACT B
CACHE:IDENTITIES

TouchDesignerPython


Description

Cache:Identities is an interactive installation exploring algorithms' influence on social identities in the digital age, emphasizing curated content's impact on individual and collective identities. Through visual narratives on cache data, it reveals how algorithms shape our identities and challenges algorithmic opacity, prompting reflection on user autonomy and agency in digital spaces.

Medium: Generative Visuals
Format: Interactive Installation
Dilemma: Algorithm Transparency






Phase 01

To initiate the process, cache data is extracted from each volunteers’ Instagram explore pages utilizing the inspect element function on Google Chrome. Subsequently, the extracted cache data is methodically organized into folders, symbolising cache databases.

These folders, each containing 20 randomly selected images, play a crucial role as primary source materials for the subsequent phases. The images are also randomly selected, without previewing to retain the integrity of the algorithmic selection process.




Phase 02

Phase 2 of the visual narrative transitions from data collection to uploading cache data to the cloud. Visually, this phase illustrates the process of transferring the downloaded cache images to a database for analysis. The interpretive significance lies in demystifying the backend operations of how algorithms store and classify their personal data.

Phase 2 also fosters transparency in the algorithmic curation data flow, encouraging viewers to contemplate the digital trajectory of their images and their integration into a larger system. Through Phase 2, viewers can better understand the connection between individual experiences and the collective data landscape, fostering deeper reflections on the curation process.




Phase 03

Phase 3 of the visual narrative aims to visually represent the expansive nature of the data stored in the cloud. The arrangement of the images and manipulation of the z-axis create an endless landscape, symbolizing the vastness of the database. Despite utilising a limited number of images, the goal is to convey the vastness of data, spanning from the inception of one’s digital presence to the present, all coexisting within the same database.

By presenting the data landscape as endless and boundless, the visuals aim to prompt viewers to contemplate the sheer volume of information perpetually stored and algorithmically processed. The intention is to provoke a deep reflection on the extent of digital footprints and the comprehensive nature of algorithmic curation, encouraging viewers to consider their online presence within this vast and intricate data landscape.




Phase 04

Phase 4 of the visual narrative introduces dynamic line visuals derived from multi-image blobtracking technique. These lines intricately connect images, symbolising the complex relationships formed between data within the algorithmic landscape. The interpretive significance lies in depicting interconnectedness and the dynamic nature of data processing. The introduction of movement imparts depth and dimensionality to the algorithmic black box while suggesting the data’s versatility, enhancing the complexity of the visual narrative.




Phase 05

Phase 5 of the visual narrative encapsulates the journey’s essence, highlighting how cache data gradually intertwines with and shapes our identities in the digital realm. Through a series of immersive visuals, viewers are prompted to delve into introspection, contemplating the profound implications of algorithmic influence on personal identity formation.

By integrating the webcam into this phase, the viewer’s engagement with the experience is heightened. The inclusion of the webcam serves to create a more interactive and immersive environment, allowing viewers to directly connect with the narrative unfolding before them.



ARTEFACT B DOCUMENTATION









ARTEFACT C
FILTER BUBBLE
INCUBATOR

TouchDesignerPython


Description

Filter Bubble Incubator is a visual exploration revolving around the formation of filter bubbles. Drawing parallels to embryos in an incubator, the project vividly illustrates how content personalisation within these bubbles evolves under distinct conditions, shaped by individual user activity history and personal information. Through this exploration, the project aims to spark reflection on the diverse perceptions of truth influenced by algorithmic personalisation.

Medium: Generative Visuals
Format: Visual Installation
Dilemma: Filter Bubbles






Ad Topics

Ads play a crucial role in shaping the personalized content experience that defines a user's filter bubble, influencing their perceptions and behaviors within their online environment. Meta Ad Topics, a feature within the Meta platform, empowers users to control the topics of ads they see, offering insight into the categories of ads targeting them.

Using this data from 12 individuals, the project visually explores the diversity of content exposure that resulted from personalisation.




Filter Bubbles

Each ad topic is categorized into one of eight groups: lifestyle, business, travel, arts, culture, fashion, sports, and general interests. I calculated the percentages of ad topics within each category from an individual's data and used them to create 3D geometric node bubbles.

These bubbles, representing each category, were then combined to create a visualisation of the individual's filter bubble.




Content Exposure

The bubbles are positioned over generative text to demonstrate the impact of our filter bubbles on the content we encounter. By combining the bubbles from each individual, I arranged them akin to eggs in an incubator to produce a comprehensive visualisation.

This final display allows viewers to perceive the distinctiveness of our individual filter bubbles and encourages contemplation about the varying perceptions of truth that result.



ARTEFACT C DOCUMENTATION