ReflectAI
AI-assisted journaling to support emotional awareness and reflective practice
Abstract
ReflectAI is a journaling platform that integrates AI to encourage emotional awareness and reflective practice. Six university students participated in a longitudinal study over 21 days, using AI-generated prompts for journaling and completing pre/post surveys with optional interviews. Findings indicate that ReflectAI reduced blank-page anxiety, supported stress processing, and fostered deeper "why/so-what" reflection. Quantitative measures showed gains in both emotional awareness (+0.34, 3.58 → 3.92) and confidence in understanding emotions (+0.42, 3.33 → 3.75), while qualitative feedback highlighted benefits in organizing thoughts, processing academic stress, and providing a supportive/non-judgmental space. This research offers early evidence for student reflection and emotional growth, motivating future work in educational contexts.
Keywords: AI, Students, Journaling, Stress, Reflection, Emotional Awareness, HCI
About the Research
Motivation
- Reduce blank-page anxiety in journaling
- Encourage deeper "why" and "so-what" reflection
- Sustain consistent reflective practice
Contributions
- Prototype AI-assisted journaling app
- Longitudinal study with n=6 participants
- Early evidence on emotional awareness and usability
Research Team
Pujan Pokharel
Research Student
Drexel University
Undergraduate researcher focused on human-computer interaction and AI-assisted tools for personal growth. Led the design, development, and evaluation of ReflectAI, conducting user studies and analyzing qualitative and quantitative data.
Dr. Tim Gorichanaz
Research Advisor
Drexel University
Associate Teaching Professor & Associate Department Head for Graduate Affairs, Information Science in the College of Computing & Informatics, specializing in information behavior, human-centered design, digital ethics, and philosophy of technology. Provided guidance on research methodology and theoretical framework.
Study Design
Some Measures Used
Frequency of journaling
System-logged (number of journals, days active)
Minutes per session
Self-reported via post-survey
Journal length
Average characters per journal, auto-logged in the app
Emotional awareness
1–5 Likert, 0.5-step (e.g., "I can identify and understand my emotions")
Confidence in emotional understanding
1–5 Likert, 0.5-step (e.g., “I’m confident I can identify and understand my emotions.”)
Usability
SUS-style items (ease of use, clarity, non-judgmental tone)
Protocol
- Onboarding + Pre-survey
- Journaling for 21 days
- Post-survey + optional interview
Participants
- Population: University students (n = 6)
- Context: Academic term / student life
- Devices: Mostly laptop and mobile/tablet
Demographics
- Population: All undergraduates (ages 21–25)
- Majors: 2 Computer Science, 2 Biology, 1 Business, 1 Data Science
- Gender balance: 3 Female, 3 Male
- Context: Students balancing coursework, exams, and part-time work
Ethics & Consent
- Informed consent, voluntary withdrawal permitted
- No clinical claims; app supports reflection, not therapy
Results
Across six participants, ReflectAI was used most frequently on laptop and phone/tablet devices, with an average of 6 sessions per participant over the study period (36 total journals). Emotional awareness increased from 3.58 → 3.92 (+0.34), and confidence in understanding emotions increased from 3.33 → 3.75 (+0.42). Participants reported that ReflectAI helped them organize thoughts, process academic stress, and provided a supportive/non-judgmental space for emotional exploration. Common improvement suggestions included enhanced visual design, easier access to past reflections, and more customizable prompts.
Session Metrics
Per-Student Breakdown
Usage Frequency
Data Analysis & Visualizations
Below are detailed charts and analysis showing the impact of ReflectAI on participants' emotional awareness, usage patterns, and qualitative feedback throughout the study period.
Group Impact — Emotional Awareness
(1–5 Likert, 0.5-step)
- Pre average: 3.58 → Post average: 3.92 (+0.34 net change)
- ReflectAI reduced blank-page anxiety and supported deeper "why/so-what" reflection.
Data source: study surveys
(1–5 Likert, 0.5-step).
Per-Participant Emotional Awareness (Pre vs Post)
- Most students improved or held steady; no declines observed.
- Individual variation reflects study schedules and journaling frequency.
Paired bars per participant — 1–5 Likert, 0.5-step.
Group Confidence — Pre vs Post
(1–5 Likert, 0.5-step)
- Group mean rose from 3.33 → 3.75 (+0.42), nudging the cohort from mid-neutral toward 'Agree'.
- Confidence gains track with journaling frequency; no declines observed.
Data source: study surveys
(1–5 Likert, 0.5-step).
Per-Participant Confidence (Pre vs Post)
- 4 of 6 participants increased; 2 held steady; no declines observed.
- Larger gains came from more engaged students; already-confident or quick-session students stayed flat.
Paired bars per participant — 1–5 Likert, 0.5-step.
Journals by Weekday (All Students)
- Wednesday was the most active day.
- Mid-week journaling aligned with assignment and work rhythms.
System-logged journals during the 21-day study.
Usage & Habits
- 36 total journals; ~6 per student on average.
- Longest streak: 5 consecutive days; median gap: 2 days.
System-logged journals during the 21-day study.
Average Journal Length by Student
- Average length ~720 chars per journal.
- Reflects short, focused sessions (see session times).
Auto-logged character counts.
Session Length Breakdown (Self-Reported)
- 42% under 10 minutes, 39% between 10–20 minutes, 19% over 20 minutes.
- Typical session ~15 minutes (self-reported).
Post-survey, self-reported.
Qualitative Themes (n of 6 participants)
- Search past journals — 4/6
- Design/UI polish — 3/6
- Easier access to past reflections — 3/6
- Customizable prompts — 1/6
- Offline mode — 1/6
- School-focused prompt packs — 1/6
Counts reflect unique suggestions captured per participant through survey and interview.
Qualitative Themes
What Participants Liked Most
-
"Clear and targeted AI prompts really helped me untangle messy thoughts — super helpful!"
-
"It felt like therapist-like guidance without judgment, which was very supportive."
-
"The ease of use and simple interface made journaling much less intimidating for me."
-
"It was useful for increasing emotional awareness and reflection, in my opinion."
-
"I think the supportive and non-judgmental tone probably helped me open up about difficult emotions."
-
"Perfect for my schedule. I could do quick sessions that fit right between classes and work."
Suggestions for Improvement
-
"A more polished, notebook-style design would make the experience more engaging for me."
-
"It’d be amazing to have easier access to past reflections, which would help me see how I’ve grown!"
-
"I think customizable AI prompts would probably make the experience feel more personal and relevant."
-
"Offline mode would be really valuable for journaling without an internet connection, in my opinion."
-
"A search function for past journals would save me time and let me quickly find the reflections I need."
-
"School-focused prompt packs would be super helpful during exams, projects, and finals week!"
Limitations
-
Small n (6 participants)
-
Homogeneous group (all undergraduates, 21–25)
-
Short duration (3 weeks)
The App
Open App
Journal Editor
Session Summary
Implementation (tech overview)
- Prototype: React + Supabase + Cohere/OpenAI
- This site: HTML + CSS + vanilla JS
- Charts: Lightweight SVG (no external libraries)
Acknowledgments
This research was conducted with guidance from Dr. Tim Gorichanaz at Drexel University College of Computing and Informatics. Special thanks to the study participants for generously sharing their time and reflections.
Conclusion
ReflectAI demonstrates the potential of AI-assisted journaling to reduce blank-page anxiety, promote deeper reflection, and provide supportive guidance for stress processing in academic life. Although the study was small in scale (n=6), participants consistently reported gains in clarity, confidence, and emotional awareness. The three key benefits—reduced blank-page anxiety, deeper reflection, and stress processing support—show promise for student populations. Future work should explore longer-term retention and well-being outcomes, diverse majors and backgrounds, and how AI can complement (not replace) human support systems like academic advising and counseling services.