07 / 09COMPUTATIONAL DESIGN · DATA

Open Space Scan

Testing where AI belongs in a responsible design workflow

RoleDesigner & developer

StatusCourse study

Open Space Scan is a desktop evidence workspace for early open-space investigation across Greater Sydney. It moves from a regional pattern to one selected area, comparison and local verification without turning incomplete spatial data into a recommendation.

This evidence workspace supports investigation order only; it does not make a planning recommendation.

Year: 2026
Context: CODE1161 Assessment 2 · Coursework
Disciplines: COMPUTATIONAL DESIGN, DATA, AI WORKFLOW
Deliverables: Evidence dashboard, Spatial data pipeline, AI process record, Verification framework

Direction

I built the interface and reproducible pipeline around explicit limits. Source definitions, missingness, sensitivity and non-causal associations stay inspectable, while AI use is documented as part of the making process rather than presented as authority.

Result

The working technical package keeps 373 SA2 areas and 14 source classes traceable through Find area, Overview, Compare and Verify. Its project-defined results support investigation order only; administrative submission and marking are not claimed here.

Evidence, not recommendation

The dashboard can expose patterns and rank complete records for investigation. It cannot establish statutory deficit, access, quality, ownership, capacity, feasibility, causality or funding priority, so the interface states exactly where its authority ends.

Open Space Scan Compare view with map, scatter plot and peer areas — Area-level association remains visible and explicitly non-causal

Open Space Scan Verify view with local checking criteria and evidence export — The workflow ends with local checks and exportable evidence

A pipeline that can be questioned

Every rendered record traces through pinned inputs, geometry reconciliation and validation checks. Sensitivity testing makes alternate definitions visible instead of presenting one configuration as inevitable.

Open Space Scan reproducible evidence pipeline — Pinned inputs to reproducible workspace

Sensitivity analysis comparing four recreation class definitions — Alternate class definitions change the result

AI as a tool, not an answer

AI helped scope alternatives, scaffold parts of the interface and pipeline, diagnose failures and plan verification. Weak directions were rejected, reversals were recorded and the final artefacts were checked against pinned evidence.

Project details

Deliverables

Evidence dashboard, Spatial data pipeline, AI process record, Verification framework

Tools

Python, JavaScript, MapLibre GL, Playwright, AI-assisted workflow

Credits

AI supported option generation, scaffolding, debugging and verification planning. It was not a planning-data source or decision maker; evidence selection, interpretation and final accountability remained mine.