Exploring and Probing the Algorithmic Gaze on Bodies and Well-being

Paper Title

Exploring and Probing the Algorithmic Gaze on Bodies and Well-being

Publication Info

• Topic area: Human-Computer Interaction (HCI) and Machine Learning (ML) in wearable self-tracking technologies.
• Keywords: Algorithmic gaze, wearable technologies, self-tracking, machine learning, autoethnography, fabulation, quantified self, design material, well-being, bodily data.

Background and Problem

• Problem / challenge: Wearable self-tracking technologies increasingly integrate ML to provide personalized classifications and recommendations, but they face challenges such as bias, exclusivity, opacity, and normative views on bodies and well-being.
• Significance: Understanding how ML models influence daily interactions with wearable devices is crucial for designing inclusive, reflective, and less prescriptive technologies that support well-being.
• Motivation and related work: Prior research critiques wearable technologies for perpetuating body normativities, inaccuracies, and privacy concerns. ML-driven wearables have been studied for their technical capabilities but lack exploration of experiential impacts and alternative design approaches.

Solution

• Proposed approach: Conceptualization of the algorithmic gaze and use of fabulation to critically investigate and reimagine ML-driven wearable self-tracking technologies.
• Novelty:
  1. Introduction of the algorithmic gaze as a conceptual frame to explore ML's influence on body and well-being.
  2. Use of fabulation to craft narratives that critique and soften the algorithmic gaze.
  3. Identification of design openings for wearable technologies, focusing on experiential qualities, interpretative approaches, and collective expressions.
• Procedure and key techniques:
  • Conducted a five-month autoethnographic study with the Oura Ring wearable, documenting daily interactions and reflections.
  • Thematic analysis to identify tensions between wearer and ML models, categorized into three themes.
  • Developed fabulations based on findings to explore alternative design spaces, accompanied by visual illustrations.

Results

• Concrete findings:
  • Three themes emerged: conflicting narratives of bodily activities, reversed reinforcement learning, and blurry boundaries of multiple bodies in continuous data streams.
  • ML models often contradicted personal perceptions, influenced vocabulary and self-perception, and raised questions about collective data entanglements.
• Advantage over baselines:
  • Proposes a shift from evaluation to interpretation, and from individualization to collective expressions, offering more reflective and inclusive design approaches.
• Experiments / evaluation:
  • Autoethnography provided detailed, situated insights into daily interactions with the Oura Ring.
  • Fabulations probed alternative ways of experiencing ML outputs, emphasizing uncertainty, ambiguity, and multiplicity.
• Limitations and future work:
  • Findings are context-specific and not generalizable to all wearables or bodies.
  • Future work should incorporate technical investigations, center marginalized perspectives, and explore fabulation in other ML contexts.

Summary

This paper critically examines the role of ML in wearable self-tracking technologies through the lens of the algorithmic gaze. A five-month autoethnographic study with the Oura Ring identified tensions between wearer and ML models, categorized into three themes. Using fabulation, the authors crafted narratives that probe these tensions and propose design openings emphasizing experiential qualities, interpretative approaches, and collective expressions. The study contributes nuanced insights into ML's impact on daily interactions and introduces fabulation as a method to explore ML as a design material, inviting further research in this space.