AI-inclusive education

AI should help students find the course—not invent the course.

The planned member assistant is grounded in approved Quant Institute material and keeps human authority visible.

What it may do

Locate the relevant lesson, assignment, rubric, or supplemental section.
Explain documented concepts and terminology.
Clarify what an assignment requires.
Offer a bounded first-move hint.
Review work a student has supplied for missing requirements or contradictions.

What it may not do

Provide personalized financial advice or buy/sell recommendations.
Complete untouched graded work.
Predict grades or claim instructor approval.
Make admissions, certification, policy, or account decisions.
Reveal instructor-only or archived sources.

Source grounding

Only material explicitly classified as public or approved student-facing content should be indexed. Answers should link to the smallest useful course section and acknowledge when the approved material does not support an answer.

Human path

Members can send unsupported, sensitive, or unresolved questions to a human. Repeated questions should improve the source material and FAQ rather than teaching the assistant to improvise.

Privacy and review

Conversation retention must be decided before launch.

The production system will disclose what conversation data is retained, why it is retained, for how long, and who may review it. Sensitive student work and unnecessary personal information should not be collected by default.

The assistant will be tested against restricted-source requests, prompt injection, financial-advice requests, graded-answer requests, conflicting content, and unsupported questions before member release.