We audited 31 leading universities across Kazakhstan, Kyrgyzstan, Uzbekistan, Tajikistan and Turkmenistan on two questions: do they publish a public AI usage policy, and what AI programs do they teach? The answers point in opposite directions.
Teaching fast, governing slow
Eighteen of the 31 universities (58%) now teach AI-related programs — machine learning, data science, applied AI tracks. Only six (19%) publish a verifiable institutional AI policy. The ratio is roughly three to one, and it is the highest teaching-to-governance gap we have measured anywhere.
This is not a story of backwardness. It is a story of speed. Central Asian universities moved fast to add AI curricula — correctly reading where student demand and labor markets were heading — but institutional governance, which requires committees, legal review, and cross-faculty consensus, moves at the speed of universities everywhere: slowly.
Country breakdown
| Country | Universities audited | Teach AI | Publish AI policy |
|---|---|---|---|
| Kazakhstan | 10 | 8 (80%) | 3 (30%) |
| Uzbekistan | 7 | 6 (86%) | 1 (14%) |
| Kyrgyzstan | 7 | 4 (57%) | 2 (29%) |
| Tajikistan | 3 | 0 (0%) | 0 (0%) |
| Turkmenistan | 4 | 0 (0%) | 0 (0%) |
Source: Central Asia AI Education Audit 2026 (31 universities). Percentages rounded.
The blind spots
Turkmenistan and Tajikistan are near-complete blind spots: almost nothing verifiable from official domains in either category. This does not mean nothing is happening — it means nothing is published. For researchers, journalists, and prospective students, unverifiable and absent are functionally identical.
Why the gap matters
When a university teaches large language models in the morning and has no policy on their use by afternoon, three things happen. Exam boards improvise. Ambitious students self-censor — or don't, and hope. And the institution's own AI researchers work under rules written for a pre-AI university.
The fix is not to slow the teaching. It is to let governance draft behind the curriculum: a one-page interim guidance document, published now and refined later, beats a perfect policy that ships in 2028. Six universities in the region have already done it. Their documents are linked in our dataset — free to copy, translate, and adapt.
Data currency check. Re-verified 2026-10-19: 31 universities, 18 teaching AI (58%), 6 with published policies (19%) — unchanged. The country breakdown (KZ 8/10 teaching, UZ 6/7, KG 4/7, TJ/TM zero) is current as of the audit date. No new policies detected in the region since publication. Tajikistan and Turkmenistan remain blind spots; if you can verify a policy from either country, send it — we publish corrections with attribution.
— Study Intelligence, from the Central Asia AI Education Audit 2026. Data and methods open under CC-BY-4.0.
Study Intelligence is a research initiative by ACLAS College.