Clones in the Machine: A Feminist Critique of Agency in Digital Cloning Conference: ACM CHI Conference on Human Factors in Computing Systems; May 2025; Yokohama, Japan

Siân Brooke · 2025

Read Clones in the Machine: A Feminist Critique of Agency in Digital Cloning Conference: ACM CHI Conference on Human Factors in Computing Systems; May 2025; Yokohama, Japan by Siân Brooke (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-18807.txt.

Introduction

Siân Brooke offers a feminist critique of digital cloning in research, in a paper from 2025. Digital clones copy a person's data to mimic their behavior, and are sold as handy tools for behavioral insight at scale. Brooke argues that this framing hides hard questions about consent, agency, and who gets represented.

Drawing on feminist theories of agency, the paper says cloning flattens human complexity into tidy data. That flattening can bake in the biases of the systems that built it. As a remedy, Brooke proposes shared data repositories that no single party controls, plus consent models that can change over time. The aim is AI practice that respects context and challenges solution-first thinking in tech research.

Catalog record

CREATORSiân Brooke
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE10 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-18807

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Clones in the Machine: A Feminist Critique of Agency in Digital Cloning Conference: ACM CHI Conference on Human Factors in Computing Systems; May 2025; Yokohama, Japan Siân Brooke Abstract . This paper critiques digital cloning in academic research, highlighting how it exemplifies AI solutionism. Digital clones, which replicate user data to simulate behavior, are often seen as scalable tools for behavioral insights. However, this framing obscures ethical concerns around consent, agency, and representation. Drawing on feminist theories of agency, the paper argues that digital cloning oversimplifies human complexity and risks perpetuating systemic biases. To address these issues, it proposes decentralized data repositories and dynamic consent models, promoting ethical, context-aware AI practices that challenge the reductionist logic of AI solutionism. 1. Introduction In September 2024, a viral post ( Ewe 2024 ) circulated on Meta platforms, claiming that reposting a disclaimer would prevent Meta from using user data to train large language models (LLMs): “Goodbye Meta AI, […] I do not give Meta or anyone else permission to use any of my personal data, profile information, or photos.” Despite its popularity, Meta confirmed the disclaimer had no legal effect ( Ewe 2024 ) , exposing the gap between perceived and actual user agency in artificial intelligence (AI) systems. Digital replicas of users captured in their data persist without consent, raising questions about data ownership and autonomy. Similar tensions surfaced in 2023, when users on Stack Overflow and Reddit 1 1 1 https://www.techzine.eu/blogs/applications/120084/openai-deal-with-reddit-shows-complete-lack-of-user-agency-over-content/ staged data deletion protests after announcements that their contributions would be used for LLM training. These protests represent refusals to participate in systems that commodify digital labor, highlighting demands for digital autonomy. At the center of these debates are digital clones—computational agents that replicate users’ digital histories (e.g., posts, interactions) to simulate and predict behaviors ( Puri et al. 2024 ) . Touted as the “Holy Grail” for understanding human behavior ( Truby and Brown 2021 ) , digital clones reflect a broader trend of AI solutionism, which assumes technical solutions can resolve complex social issues. For example, Schmidt et al. ( Schmidt et al. 2024 ) propose how LLMs can generate personas that mimic human characteristics and respond to surveys in Human-Computer Interaction (HCI) research, promising scalable insights without involving real participants. However, this framing obscures significant ethical concerns. Digital clones do more than aggregate data—they simulate behaviors and agency, risking the reduction of complex human identities to predictive models. Assumptions of representational accuracy overlook relational and contextual dynamics that shape behavior. These issues are particularly pressing when viewed through feminist theories of agency ( Schwartz and Neff 2019 ; Toupin 2024 ) , which emphasize that autonomy is relational and shaped by power structures. This position paper examines these ethical challenges, focusing specifically on researcher-driven digital cloning in simulation studies and exploring how feminist perspectives can guide more ethical, context-aware practices in AI research. 2. Digital Clones in Research Digital clones replicate user-generated data to simulate behaviors and interactions. These clones emerge through two primary approaches: (1) User-Driven Cloning : Users voluntarily contribute data for research, providing explicit consent. (2) Researcher-Driven Cloning : Researchers scrape and replicate user data, typically from public platforms, without explicit consent. I focus on researcher-driven digital clones used in simulation studies, where researchers replicate users’ digital histories to model social network dynamics, such as misinformation spread and the formation of political echo chambers. For instance, Puri et al. ( Puri et al. 2024 ) developed a simulation framework combining agent-based modeling and natural language processing (NLP) to clone a misinformation-sharing network. The study downloaded histories of over 10,000 users, reconstructed the network, and manipulated algorithms to observe interaction shifts. While these simulations offer valuable insights into polarization and information flow, they raise significant ethical concerns. Although Puri et al. ( Puri et al. 2024 ) obtained ethical approval, the ethics board deemed explicit consent unnecessary due to the data’s public availability. However, this overlooks the critical distinction between publicly available data and a digital replica. Unlike passive data collection, digital clones actively simulate user agency, predicting behaviors and interactions. Assuming these clones represent human complexity overlooks relational factors shaping behavior, such as exclusion or disengagement based on gender or ethnic identity. Additionally, treating public data as freely usable without consent undermines privacy and autonomy ( Fiesler 2020 ) , echoing exploitative practices like Cambridge Analytica. 3. Feminist Critique of Digital Cloning Feminist academics such as Toupin ( Toupin 2024 ) , Suchman ( Suchman 2020 ) , and Nunes ( Nunes 2024 ) highlight concerns about agency loss in AI deployment. Particularly compelling is Suchman’s ( Suchman 2020 ) critique of binary oppositions, such as subject/object and human/machine, which often privilege one element over another. These dichotomies obscure the relational nature of agency ( Suchman 2020 ) , shaping how digital clones are understood. Rather than being passive representations, the data generated by digital clones functions as an extension of the agency created by users. This perspective underscores the need for researchers to recognize the interconnectedness of humans and machines, as agency does not reside in discrete entities but instead emerges from sociomaterial arrangements. Digital clones, therefore, must be seen as entangled within broader configurations of algorithms, datasets, and (biased) human decisions. 4. Ethical Concerns in Digital Cloning Emerging feminist critiques stated above highlight ethical concerns in digital cloning, particularly around agency, representation, and power. Key assumptions in digital cloning research include: (1) Clones sufficiently capture individual complexity; (2) Anonymized data removes the need for consent. However, these assumptions obscure the relational and contextual nature of user identities. Drawing on Suchman’s ( Suchman 2020 ) argument, it becomes clear that digital clones risk flattening nuanced interactions. For example, personal relational choices—such as expressing political opinions differently in private groups versus public threads—can shape social dynamics. Detached from these contexts, digital clones may misrepresent behaviors, treating all interactions as uniform. Similarly, clones trained on past engagement patterns may overemphasize frequently repeated behaviors while overlooking passive or evolving user interactions, reinforcing echo chambers and distorting online discourse. Furthermore, as Toupin ( Toupin 2024 ) emphasizes, digital footprints are extensions of the self, especially for marginalized communities. Replicating these selves without explicit consent challenges user agency and risks reinforcing inequalities. Persistent clones, even after data deletion requests, deny users the right to be forgotten. Reinforcing Inequalities through Cloning: AI-generated deepfakes show how identity replication can exploit marginalized groups ( Karnouskos 2020 ) . Similarly, digital clones can misrepresent or erase diverse voices, reproducing systemic biases. The historical case of Henrietta Lacks reminds us that data extraction without consent—biological or digital—disproportionately harms oppressed groups. Feminist perspectives demand recognizing that consent is a continuous negotiation shaped by power dynamics. Behavioral Manipulation and Profiling: Nunes’ ( Nunes 2024 ) posthuman feminist framework highlights how AI systems shape user behavior. Digital clones can enable behavioral manipulation through algorithmic nudges, normalizing subtle control forms. Predictive profiling risks reinforcing biases, marginalizing vulnerable populations. When clones persist despite deletion requests, users lose control over their representations, raising questions about data sovereignty. 5. Re-imagining Ethical Frameworks for Digital Cloning Decentralized Data Donation Repositories: To address data power imbalances, decentralized, ethically governed repositories are essential. These repositories would enable voluntary, community-driven contributions under transparent

Read the complete text