Working Group Charter: Islam, AI & Education

Last updated 31 July 2026

Group name

Islam, AI & Education

Mission

Build an adaptive learning platform for the Islamic education of 13–18 year olds — an established curriculum made adaptive, with mastery tracking and AI tutoring, grounded in the tradition.

Deliverable

An adaptive learning platform built properly, end-to-end: an existing Islamic-studies curriculum made adaptive — mastery tracking, adaptive sequencing, and AI tutoring on an open-source foundation. Then, as the second stage, a pilot with a real cohort: a weekend school or two from the network.

In scope

• Selecting an established Islamic-studies curriculum and making it adaptive — the work is the adaptation, not writing a curriculum from scratch.
• The platform, built end-to-end on the research’s recommended foundation — a fork of OATutor (MIT-licensed adaptive tutoring from UC Berkeley: mastery tracking, adaptive problem selection, content authored through its spreadsheet-to-JSON pipeline).
• Scholarly review of content and AI-generated tutoring before students see it, with the tutor layer grounded in vetted Islamic sources.
• Safeguarding from day one: 13–18 means minors — privacy, parental consent, and safeguarding requirements are part of the spec, not discovered later.
• The pilot: real students at a weekend school or two, and iterating from what the pilot finds.

Out of scope

• Building an LMS. We will just use OATutor — the group’s effort goes into the educational experience, not platform plumbing.
• A safe-adoption framework for AI in Islamic schools — a separate project, not this group’s deliverable.
• Age bands outside 13–18. The research notes options for younger children (e.g. Oppia); revisited after the pilot proves the model.
• Free-form Islamic question-answering. That is Ansari’s domain — the platform’s conversational tutor builds on that work rather than duplicating it.
• The commercial-platform route (Sana Learn, ~$47k/yr floor). It stays the researched fallback if speed ever matters more than control — a deliberate reserve, not a parallel track.

Duration, timeline, and milestones

Choose the curriculum, build the platform end-to-end, then pilot it: select an established curriculum and map it onto the adaptive engine, build the full platform properly, and put it in front of real students at a weekend school or two as the second stage.

T₀ = the chartering date (co-leads confirmed). Offsets firm up at the first milestone review.

Resources needed

• Scholarly review of content and AI tutoring before it reaches students.
• Curriculum partnership: rights or permission to adapt the chosen curriculum.
• Pilot partners: introductions to one or two weekend schools from the IASER network.
• Modest hosting and inference: the open-source foundation deploys without licensing costs; LLM tutoring inference is gated and small at pilot scale.
• From IASER: platform and infrastructure support, and editorial help turning the pilot findings into a public write-up.

Audience

Students aged 13–18 in Islamic and weekend schools, and self-directed teen learners; the teachers and schools who run the platform with them; and the parents whose consent and confidence the safeguarding work earns.

How we work

The work lives in a new public repo under github.com/iaser-ai (created at chartering), including the curriculum spec and the content pipeline. Meeting cadence set by the co-leads at chartering.