About AAII — story, mission and roadmap

Our story

The best free courses in the world are not free for everyone

The excellent open AI curricula all quietly assume four things: that you read English fluently, that you have a stable connection, that you own a machine which can run a notebook, and that there is someone to ask when your code breaks at midnight.

Remove any one of those and the course becomes difficult. Remove all four and it was never really available to you. Meanwhile the mathematics is not the hard part — motivated seventeen-year-olds in Kabul learn calculus from a photocopied book. The hard part is that nobody built the road.

So we started from the other end. We asked what an AI education would look like if it were designed for a student in Herat with an Android phone, a connection that drops twice an hour, and nobody nearby to ask — and then made sure it was also excellent for a graduate in London or Toronto. Designing for the harder case produced a better course for everyone.

AAII is that course, plus the community that makes finishing it possible. We are small, we are open about what we do not yet have, and everything we make stays free.

Mission and values

Teach artificial intelligence so thoroughly that our students no longer need us

These are the rules we hold ourselves to. They are specific on purpose, so you can tell when we break one.

Free, and free means free

No fees, no locked modules, no "premium" tier, no advertising, and no selling learner data. If we ever cannot afford to run something, we will say so rather than charge for it.

Teach the idea, not the tool

Libraries change every eighteen months. We teach why a method works so that the next framework takes you a weekend, not a year.

Plain language, precise terms

Nothing is dumbed down and nothing hides behind vocabulary. When we use a technical term we define it once, properly, and then use it consistently.

Show the failures

We teach where models break, who they harm when they break, and how to measure it. A course that only shows working code produces engineers who cannot debug.

Local data, local problems

Wheat rust, local speech data, clinic queues, earthquake damage, textbook OCR. Learning on problems that matter to you is faster and the result is worth keeping.

Open by default

Curriculum, slides, code, datasets and our finances are published. Anyone may translate our material or teach from it, including in a classroom we will never see.

Why AI education matters here

A country with no builders becomes a permanent customer

Artificial intelligence is turning into infrastructure — the way electricity and mobile networks did. Infrastructure gets built by whoever shows up with the skills. Places that only ever buy the finished product pay forever, adapt slowly, and have no say in how the thing works.

Language technology is the clearest case. Speech recognition for the languages spoken here will exist only if people who speak them build it. No global lab has a commercial reason to prioritise a language with no advertising market. The same goes for optical character recognition on Afghan school textbooks, and for a medical assistant that works in the language a patient actually uses. These are not charity projects; they are open engineering problems with no owner.

And the work is portable. A data or machine-learning job can be done from anywhere with a laptop and electricity. For young Afghans — women especially, for whom leaving the house to attend a class may not be possible — a remote technical skill is one of the few routes to an independent income that does not require permission or a plane ticket.

We are not claiming AI fixes anything by itself. We are claiming that being able to build it is better than not.

Where we are going

Four phases, in order

This is a sequence, not a menu — each phase depends on the one before it. We are honest about which phase we are actually in.

Phase 1 · now

AI education programmes

Ten complete subjects, from Python and mathematics through to agentic AI and MLOps, with video, written guides, code and a project in each.

Phase 2 · next

Technology development

A small research group on Afghan language technology: open datasets, benchmarks and tools that do not exist yet and that only we are motivated to build.

Phase 3

Students and entrepreneurs

Mentorship pairs, an equipment fund for learners without a laptop, and support for graduates turning a project into a product or a paid contract.

Phase 4

Opportunity through digital skills

An employability track — portfolios, interviews, remote-work practice — and partnerships with employers who will actually hire from it.

Team

Who is doing this

A small volunteer team, and room for more. If you can teach, translate, or review a lesson, there is a place for you.

Fraidoon Omarzai

Co-founder · MSc Artificial Intelligence

Jalaluddin Obaidi

Co-founder · International Business Management MSc

Transparency and impact

What we publish, and how we count

Numbers are easy to inflate, so here is exactly how ours are measured. Figures are updated at the end of each quarter.

—Learners who finished at least one course

—Lessons published

—Volunteer hours contributed

—Open datasets released

How AAII measures and reports its work
What we report	How it is measured	Published
Course completions	A learner who submits the final project of a course. Video views are not counted as learning.	Quarterly
Reach and completion	Lessons watched or read, and how many learners reach the end of a subject. We report drop-off honestly rather than only the headline figure.	Quarterly
Income and spending	Every donation and every cost, itemised. Names of individual donors are withheld unless they ask otherwise.	Annually
Curriculum licence	Lessons and slides under Creative Commons BY-SA. Code under MIT. You may teach from all of it commercially.	Continuous
Dataset provenance	Where the data came from, who consented to its use, and the licence it carries. Datasets without clear consent are not released.	Per release
Learner privacy	Progress is stored in the learner's own browser. We do not require an account to study and we do not sell anything to anyone.	Continuous

Formal non-profit registration is in progress. Until it completes we do not solicit public donations — see Get involved for the current position.