Human–AI Interaction / Exploratory Research
Experimenting with Generative AI to Reflect
When AI guides reflection, who directs the experience?
Five exploratory investigations examine AI guidance through yoga, voice conversation, deliberate compliance and resistance, emotional reflection, and dialogue with a friend. As an undergraduate researcher in Jennifer Turns’s directed research group, Jana used self-observation to ask where useful guidance ends and self-direction begins.
Five investigations into AI guidance, reflection, embodied experience, and autonomy. Original project diagram.
Year
2025
Institution
University of Washington, Human Centered Design & Engineering
Jana’s role
Undergraduate researcher
Adviser / collaborators
Directed research group with Jennifer Turns
Why this research
AI can offer a direction for reflection, but the usefulness of that direction does not settle who should guide the experience. This project brings that tension into everyday activities, where bodily attention, personal interpretation, and interpersonal responsibility complicate the idea of a helpful recommendation.
Approach and scope
Five exploratory, self-observational investigations were conducted in Jennifer Turns’s directed research group. The settings include yoga, voice conversation, deliberate compliance and resistance, emotional reflection, and conversation with a friend. These are situated inquiries into experience, not a controlled comparison or a representative study.
What this work contributes
The work establishes five settings for examining the boundary between guidance and self-direction. Its contribution is exploratory: it turns a general concern about AI autonomy into specific questions that can be pursued through original prompts, contemporaneous notes, and appropriately shared conversation records.
Following a voice, attending to a body
Pending · Yoga notebook / original activity photograph
What happens between an instruction and a sensation?
Pending · Voice transcript excerpt
Conversation as a reflective setting
Voice-based conversation became a setting for personal reflection, bringing an external response into a self-directed activity.
When does a response support reflection—and when does it redirect it?
Following, resisting, leaving it to chance
Deliberate compliance, resistance, and randomness provided three ways to respond to AI guidance.
What makes following a recommendation feel chosen?
Pending · Original prompt
Pending · Response + dated observation
Who gives an experience its meaning?
Pending · Redacted reflection notebook
Who determines the direction of a reflective exchange?
A third presence in a conversation
AI-mediated dialogue with a friend brought the question of autonomy into an interpersonal setting. What was disclosed about AI involvement matters to how the exchange is understood.
What responsibilities arise when AI enters communication between people?
Pending · Consented, anonymized dialogue excerpt
Reading across the work
The investigations can be read through three questions: who initiates a direction, how a person accepts or resists it, and who interprets the resulting experience. These questions connect embodied activity to conversational and interpersonal settings. They remain a proposed synthesis until supported with dated records; no recurring behavioral effect is claimed.
From guidance to self-direction
These are exploratory, self-observational investigations, not causal or population-level findings. Disclosure, consent, privacy, and the ethics of mediated conversation remain unresolved. Related questions about guidance and control later moved into physical interaction in Domesticated.
Documentation URL pending