AI-Supported Reflective Learning: First Cohort — Research Update

Cohort Overview

The first cohort of the AI-supported reflective learning study began in January 2026. Participants (n = 34) are graduate students in education and psychology programs.

The intervention uses a structured weekly reflective writing protocol with AI-generated feedback based on the PersonaMatrix engine.

Week 1–4 Observations

Engagement: 29/34 participants (85%) completed all four weekly reflections. Average response length increased from 187 words (Week 1) to 312 words (Week 4), suggesting growing comfort with the format.

Qualitative themes (preliminary, from researcher notes):

Participants initially framed reflections as performance tasks ("what should I say?") rather than genuine inquiry

By Week 3, approximately 60% showed evidence of genuine self-questioning in their writing

AI feedback was perceived as "less judgmental" than peer feedback by 22/29 respondents (Week 4 survey)

Unexpected finding: Three participants explicitly noted that the AI's questions helped them notice assumptions they hadn't previously articulated. This aligns with the theoretical model but was more prominent than expected.

Limitations

This is an uncontrolled observational phase. No comparison group. Researcher presence effects cannot be ruled out.

Next Phase

Controlled comparison begins in March 2026 with a matched cohort receiving standard peer-feedback protocol.