Choose what the system may see.
Camera, room audio, gaze tracking, room scan, browser lockdown, and flagging sensitivity create the policy input.
Project archive04 · Responsible AI / Interactive experience
Interactive ethics prototype · Illustrative classroom data
WATCHLINE
AI Proctoring Ethics Lab
An interactive experiment that makes the hidden trade-offs of AI proctoring visible: detection, false flags, privacy, and the need for human judgment move together.
Enter the live lab
Design question
Detection is not the same as justice.
Proctoring tools are often presented as neutral accuracy systems. WATCHLINE reframes the experience around the person being watched and asks what each increase in automated detection costs in privacy, context, and procedural fairness.
The interface does not deliver a correct policy. It lets visitors configure one, observe the linked consequences, inspect an ambiguous exam event, and return the final decision to people.
Experience model
Input, consequence, human decision.
Read linked outcomes, not one score.
Detection rises alongside false flags and privacy cost, preventing a single performance number from hiding the rest of the system.
Put context and appeal back in the loop.
Each simulated flag resolves through human context, then asks the visitor to approve, revise, or reject the policy.
Live ethics readout
One balanced policy, three simultaneous truths.
Simulated values that show the direction of each trade-off.
Interaction decisions
The argument is carried by the interface.
Two perspectives
Student and school views change the framing without changing the underlying evidence, exposing how institutional position shapes the same policy.
Trade-offs stay coupled
Every permission and sensitivity adjustment updates detection, false flags, privacy, policy language, and the visual surveillance state together.
Human review is structural
Ambiguous behaviour is never converted directly into punishment. The event timeline separates observation, automated flag, and human resolution.
Real interface evidence
The simulator makes policy feel consequential.
The eye, surveillance console, and evidence timeline are not decorative dashboards. They respond to policy state and help visitors connect an abstract governance choice to a person under review.
Technical foundation
Small stack, deliberate system.
- Interface
- Semantic HTML · responsive CSS · vanilla JavaScript
- State model
- Five permissions · sensitivity control · deterministic linked metrics
- Accessibility
- Native controls · visible focus · keyboard operation · reduced-motion handling
- Localization
- Complete English / Chinese interface with persistent language choice
Prototype boundary
An educational model—not a deployed surveillance product.
- The page never accesses a camera, microphone, browser history, or personal data.
- Students, exam events, percentages, and policy outcomes are illustrative simulation data.
- The formulas demonstrate relationships rather than validated predictive performance.
- The experience is a design and ethics prototype, not a school policy recommendation.
Interactive prototype
Configure the policy yourself.
Switch perspectives, change monitoring permissions, simulate ambiguous exam events, compare policy presets, and choose your position. The complete bilingual experiment runs directly in the browser.