Milestones
A long-term project, growing alongside its researcher.
MindLens Lab didn't start as a polished research program. It started with one high-school student's curiosity about how AI tools handle emotion — and a hunch that the existing approach was missing something. The depth of the question grew as the work grew. The plan below treats this as a multi-year project that evolves with its researcher: through high school, through college, and beyond. Dates are approximate — research happens in seasons, not days.
Phase 0 — The origin
How we got here.
2024 Spring
✅ DoneFirst brainstorming
Initial exploration of how AI tools might support autistic learners with emotion recognition.
2024 Fall
✅ DoneResearch + paper writing begins
Literature review on AI-based emotional intelligence training. Drafting and revisions.
2025 Spring
✅ DonePaper completed
Final manuscript submitted to Curieux Academic Journal.
2025 July
✅ Done📄 Paper published
"Artificial Intelligence in Emotional Intelligence Training for Autism" — Curieux Academic Journal.
“Seeing the paper accepted was the moment I realised this wasn't going to be a one-off project. The questions I asked at the end of the paper were bigger than the paper itself.”
Phase 0.5 — The pivot
Why dataset, not tool.
2025 Late
✅ DonePhase 2 prototyping brainstorm
Sketches of follow-up tools — AR overlays, conversational coaches, classroom modules.
2026 Early
✅ DonePivot to the dataset
Realisation: existing emotion datasets assume one correct label per moment. Building tools on that foundation would inherit the assumption. Decision: collect plural human readings as the foundational data, before any tools.
“I almost started building the app first. The pivot felt like going backwards — but every emotion dataset I looked at had the same blind spot, so building on top of them would have inherited the blind spot too. I decided I'd rather build the dataset I wished existed.”
2026 Spring
✅ DoneLab design
Locked taxonomy (9 emotions × 9 cues), schema (18 tables, 40 RLS policies), participant flow, screen inventory, hi-fi mockups.
Phase 1 — Plural reading dataset
Where we are now.
2026 May
✅ Done🚀 Production launch
mindlenslab.org is live: full participant flow, admin tools, custom domain, magic-link auth.
“Launching between AP exams was not on the plan. But once the design and the schema were locked there was nothing left to wait for, and I'd rather have real readings to learn from than another month of polish.”
2026 mid
🔄 In progressWave 1 pilot
First curated wave of clips. Target: ~50 participants across age groups and countries. Iterate UX based on feedback.
2026–2027
📅 NextWave expansion
More waves, more participants, multi-country reach. Add Korean/Chinese language support.
2027
📅 NextFirst findings paper
Phase 1 dataset analysis. Submitted as a follow-up to the Curieux paper, this time as a peer-reviewed empirical contribution.
Phase 2 — Validated study materials
After college begins.
2027 Fall+
📅 NextMethodology meets pedagogy
Use the Phase 1 dataset to design classroom exercises and self-paced learning modules that explicitly teach the plurality of emotion. Co-design with educators.
2028+
🌱 FutureEducator partnerships
Pilot the materials in classrooms serving neurodiverse learners. Iterate based on real teaching contexts.
Phase 3 — Adaptive tools
Long-term aspiration.
2028+
🌱 FutureInteractive practice tools
Tools that adapt to an individual learner — not predicting "the right answer", but offering practice in noticing variance.
Long-term
🌱 FutureWearable / real-time cueing
Possibility, not promise. Only ever grounded in validated data and shaped by the people the tools are for.