Qualia Drive

Inspiration

Traditional road-safety tests usually tell learners whether an answer is right or wrong. They rarely examine how the decision was made, whether the learner noticed hidden danger, trusted legal priority too much, reacted to social pressure, or could transfer a safety principle into a changed situation.

Qualia Drive was created to turn assessment into guided preparation for the unpredictable road.

What it does

Qualia Drive is an adaptive road-safety Trainer built with Codex and GPT-5.6.

A session contains ten scenarios. The first five create a structured learning path: Foundation, Situational Scan, Rule Violation, Social Pressure, and Transfer. The next five are selected adaptively from the learner’s weakest or least-confirmed dimensions.

After each answer, the Trainer separates legal correctness from real-world safety, examines confidence and explanation, updates an evidence-based learner model, forms diagnostic hypotheses, selects a teaching tool, and offers corrective or reinforcing advice.

All learner dimensions begin at 50% and change only through observed evidence. The system tracks attention, principle understanding, decision safety, hazard detection, confidence calibration, resistance to pressure, self-correction, knowledge transfer, and development of the Internal Trainer.

The final Trainer’s Conclusion is not a simple score. It summarizes the journey through all ten situations, identifies demonstrated strengths and unresolved safety risks, reports evidence reliability, and recommends the next training route.

How we built it

The MVP was implemented as a Next.js and TypeScript application. Codex was the primary implementation environment, used to structure the Trainer Engine, build deterministic scoring and adaptive scenario selection, create the bilingual English/Russian interface, write tests, and verify production builds.

GPT-5.6 was used to develop the educational reasoning model, including the separation of legal correctness from actual safety, competing diagnostic hypotheses, learner dimensions, changed-context scenarios, Trainer feedback, and the final conclusion.

What we learned

A correct answer is not always stable knowledge. Confidence, explanation, independent transfer, reaction to violations, and performance under pressure provide different kinds of evidence. Qualia Drive therefore treats mastery as repeated demonstrated readiness across changed contexts rather than a single correct response.

Current MVP

The current version provides a complete local ten-scenario training session, adaptive verification, bilingual localization, a compact and expanded learner profile, Trainer Advice, and a final evidence-based conclusion.

What’s next

Future versions can expand the scenario library, support verified jurisdiction-specific traffic rules, add instructor dashboards, and preserve a learner model across multiple training sessions.

Built With

• codex
• css
• gpt-5.6
• next.js
• node.js
• react
• svg
• typescript