Multilingual | STT | Prompt Behavior | QA-driven Observations
This repo collects selected user-centered reflections, bug reports, and UX notes based on my independent interaction with ChatGPT (mostly Android & Desktop, 2024–2025).
Focus areas include:
- STT issues in Korean (loss, mismatch, no edit mode, etc.)
- Prompt ambiguity and context handling
- Multilingual behavior and UI/UX friction
📁 Public Reports:
bit.ly/gpt-repo
📁 Currently working on private repo / full logs: Available on request after 30 apr.
These are not official OpenAI documents — they reflect my personal observations as a QA-linguist and advanced user.
Because user voices truly matter — especially those speaking in multiple languages. Because language goes beyond mere accuracy — it encompasses lived experience. And because these insights deserved to be documented. “While others were experimenting, my input was impactful.” As more than just a test subject, my interactions directly shaped the environment of the experiment itself. This portfolio showcases real-time interactions and thoughtful feedback provided to a leading LLM system (ChatGPT), where my input was subsequently reflected in the model's live behavior and adjustments to its dialogue tone. My linguistic input, structural suggestions, and user experience critiques were not only heard — they were visibly integrated.
| Entry | Title | Date |
|---|---|---|
| 001 | Korean STT UX Feedback & Improvement Suggestions | 2025-04-12 |
| 002 | Korean STT UX Feedback & Improvement Suggestions(2) | 2025-04-12 |
| 003 | UX Observation: Misaligned Feedback Timing in Code Execution Flow | 2025-04-12 |
| 004 | Meta-UX: Invisible Feedback Loops and Trust Friction in LLM Systems | 2025-04-12 |