Learning Systems versus Future Everyday Domestic Life: A Designer's Interpretation of Social Practice Imaginaries

Emilia Viaene, Lenneke Kuijer, Mathias FunkFrontiers Artif. Intell 2021 · DOI

Introduction

Smart home technologies (SHT) are increasingly marketed as seamless assistants that anticipate human needs through machine learning. However, a fundamental tension exists between the algorithmic requirement for predictability and the inherent "messiness" of domestic life. This paper investigates these "frictions" using a design-led forecasting method to identify how AI learning systems may fail or impose rigid structures on inhabitants during non-routine events.

Methodology: Social Practice Imaginaries

The research employs Social Practice Imaginaries (SPIs), a method that shifts focus from individual gadgets to the "doings and sayings" of everyday life. The authors selected three core domestic practices—waking up, grocery shopping, and climate control—and developed scenarios for the year 2030.

Each practice was analyzed through two contrasting lenses:

  1. The "Happy Path": A routine where the system flawlessly supports the inhabitant.
  2. The "Crisis of Routine": A scenario where the routine is disrupted (e.g., illness, unexpected guests), revealing the system's limitations.

These scenarios were further scrutinized from three perspectives: the Inhabitant (subjective experience), the Learning System (sensor inputs and logic), and the Designer (the actor whose early decisions shape system behavior).

A 'Happy Path' scenario for a morning routine, contrasting the inhabitant's narrative with the learning system's logic.

Key Insights: The "Learnability" Trap and Co-performance

A significant finding is the "learnability" trap: the risk that inhabitants will subconsciously modify their behavior to be more predictable for the system, effectively serving the machine to ensure it functions correctly. This highlights the role of SHTs as "co-performers" rather than passive tools. When a system automates a task like making coffee or opening blinds, it actively participates in the social practice, which can lead to friction if the system's logic clashes with the inhabitant's immediate needs.

The research also identifies the Designer as a "Third Actor." Frictions often stem from design-time assumptions about "appropriate" behavior or "normal" routines that do not account for the diversity of real-world contexts.

Results: A Typology of Implications

The study culminates in a typology of identified implications for designers, categorized into four domains:

  • Technological Limitations: Hardware failures or the requirement for "human preparations" (e.g., clearing floors for a robot vacuum).
  • Communication: The difficulty of systems requesting feedback without being intrusive, especially when practices are inter-related.
  • Agency: The negotiation of control. Should a system proactively "nudge" a user toward better habits, or is this an infringement on autonomy?
  • Data Usage and Safety: Beyond privacy, this includes the risk of systems being used as tools for domestic abuse or harassment (e.g., remote control of home environments by unauthorized parties).

Conclusion

Designing for the "happy path" of domestic life ignores the reality of human existence. By proactively identifying "crises of routine" through Social Practice Imaginaries, designers can better anticipate technical and social frictions. This approach encourages the development of systems that are resilient to the unpredictability of life, rather than systems that require humans to become more machine-like.

Summary updated

BibTeX
@article{viaene2021learning,
  author = {Emilia Viaene and Lenneke Kuijer and Mathias Funk},
  title = {{Learning Systems versus Future Everyday Domestic Life: A Designer's Interpretation of Social Practice Imaginaries}},
  journal = {Frontiers Artif. Intell},
  year = {2021},
  doi = {10.3389/frai.2021.707562},
}
Figure