Where Rights Show Up: The GIRAI 2023 Human Rights and AI Dimension Across Central Asia
The Global Index on Responsible AI (GIRAI) 1st Edition assessed countries on whether national frameworks, government actions, and non-state actors had produced documented evidence in twelve thematic areas at the intersection of AI and human rights. Across the five Central Asian states, the Human Rights and AI dimension is where the regional picture is most uneven — and where the gaps are most revealing. The full GIRAI thematic scores for each country are available at global-index.ai. This post reads the dimension comparatively, focusing on four areas where the regional pattern is most instructive: Education, Cultural and Linguistic Diversity, Data Protection, and Bias and Unfair Discrimination.
Education: The One Thematic Area Everyone Has Something
Education is the thematic area that comes closest to a regional constant. Every country in Central Asia has some documented activity at the intersection of AI and education — though the depth and character of that activity varies considerably.
Kyrgyzstan has the broadest coverage: framework, government action, civil society, and academic engagement all confirmed, making it the most thoroughly documented Education response in the region. National education policy includes AI literacy provisions, government has taken action on digital skills, civil society organizations have engaged, and academic institutions have incorporated AI into both research and curriculum. This breadth reflects Kyrgyzstan’s position as the country where the rights dimension was most thoroughly documented overall.
Kazakhstan’s Education dimension runs primarily through its research institutions. The Ministry of Science and Higher Education identifies 24 universities and research centers active in AI work. Four institutions — Eurasian National University (L.N. Gumilyev), Al-Farabi Kazakh National University, Satpayev University, and Nazarbayev University — are identified as the current backbone of Kazakhstan’s AI research capacity. The demographic pipeline supports this: more than 6.2 million people under 35, with 39.3% holding higher education degrees. The limitation is that this capacity is concentrated in research and technical development; the link between AI education and rights-protective AI practice is not yet institutionalized.
Tajikistan’s Education evidence runs through the AI Council’s awareness program. The Council organized awareness events at universities — including a November 2021 masterclass at MSU Tajikistan delivered by AI Council chair Azizjon Azimi — and several universities have introduced AI courses. The educational pipeline feeds directly into zypl.ai’s workforce; the content of that education does not yet include AI ethics or responsible use. The AI Council has acknowledged that ethics is part of the eventual agenda — but that acknowledgment has not produced any formal instrument.
Uzbekistan and Turkmenistan both have AI educational activity at university level. In Uzbekistan, this is connected to the 2021 presidential programme’s emphasis on creating a domestic AI innovation ecosystem. In Turkmenistan, AI programme components exist at some universities — confirmed by the GIRAI assessment — but without any ethics or responsible use dimension, and without the connecting tissue of policy dialogue, research publication, or international engagement that would make them part of a coherent national capacity.
Cultural and Linguistic Diversity: One Country’s Distinctive Strength
Cultural and Linguistic Diversity is the thematic area where Kyrgyzstan’s profile is most distinctive — and where it is most clearly doing something that no other country in the region is.
Two binding national policy documents explicitly name AI in relation to the Kyrgyz language. The Action Plan of the Cabinet of Ministers for the implementation of the National Development Program until 2026 (December 2021) — a government-wide implementation instrument covering the full National Development Program until 2026 — includes Section 597, titled «Software product development (artificial intelligence),» which assigns development of AI-based translation and semantic analysis software for the Kyrgyz language to the National Commission for State Language and Language Policy under the President. The measure is binding and covers the design and development stage of the AI lifecycle. The Program for the Development of the State Language and Improvement of Language Policy 2021–2025 (Resolution No. 51, October 2020) — adopted under the Constitution and the Law on the State Language of the Kyrgyz Republic — contains four measures specifically addressing AI and natural language processing for the Kyrgyz language in Chapter 5 («Digitalization of the State Language»), with binding scope across all state and non-state sectors. Private sector and academic actors were also found engaged on this thematic area at the time of the assessment.
The other four countries in the region had no confirmed evidence in the Cultural and Linguistic Diversity thematic area at the time of the GIRAI assessment. Kazakhstan, Uzbekistan, and Tajikistan have national languages with their own NLP challenges — Kazakh, Uzbek, and Tajik are all under-resourced in global AI training datasets — but no documented policy or institutional activity on AI and linguistic preservation or development met the GIRAI evidence standard during the research period. Turkmenistan had no evidence across any dimension.
This is not a commentary on the absence of language AI research in these countries. In Kazakhstan specifically, Nazarbayev University’s Institute of Smart Systems and Artificial Intelligence (ISSAI) has produced substantial research contributions to Kazakh-language natural language processing, including the KazLLM large language model for the Kazakh language — work that KG Labs regards as a significant contribution to Kazakh and Turkic language AI research and to the under-resourced language problem more broadly. What the GIRAI framework captures is documented policy and governance action by identified actor categories; ISSAI’s research output had not been connected to a national policy framework or government initiative of the kind the index’s evidence standard requires. The research exists; the governance bridge from it does not yet.
Data Protection: A Regional Gap
Data Protection is blank for every country in Central Asia in the GIRAI 2023 assessment — no national framework, no government action, no private sector or civil society or academic activity meeting the index’s evidentiary standard in relation to AI-specific data protection.
This does not mean personal data protection legislation is absent. Kyrgyzstan, Kazakhstan, Uzbekistan, and Tajikistan all have some form of data protection law on the books. What is absent, in each case, is alignment between that existing legislation and the requirements that AI-processed personal data creates: automated decision-making rights, algorithmic transparency, data minimization in AI training pipelines, and the specific protections that apply when data is used to build or operate AI systems. The GIRAI researcher for Uzbekistan identifies this explicitly as a gap — existing frameworks do not align with international standards for AI-processed personal data. For Kyrgyzstan, the Data Protection area was one of only four thematic areas returning zero confirmed evidence across all actor categories. Turkmenistan has a personal data protection law, but no documented implementation that would connect it to AI applications.
This regional blank is significant partly because of what surrounds it. Each country in Central Asia is investing in AI deployment in exactly the sectors where data protection matters most — healthcare AI in Kazakhstan, AI-powered loan underwriting in Tajikistan, e-government automation in Uzbekistan, public service chatbots planned in Kyrgyzstan. The data protection layer is missing from the responsible AI picture at precisely the moment when AI is being deployed at scale in domains with high concentrations of personal data.
Bias and Unfair Discrimination: Not a Conversation Yet
The Bias and Unfair Discrimination thematic area returned zero confirmed evidence across all five countries and all actor categories. No national law, no government initiative, no private sector program, no civil society activity, and no academic work addressing bias and discrimination in AI systems met the GIRAI documentation standard in any of the five assessed countries.
For Kyrgyzstan, the GIRAI researcher notes this explicitly as the most notable absence in the Human Rights and AI dimension: a country that shows evidence in eight of twelve thematic areas, and zero in this one. The pattern repeats across the region. The conversation about algorithmic bias — who AI systems work for, who they fail, what the data and design choices embedded in them reflect — has not yet become a documented subject of policy, advocacy, or academic inquiry anywhere in Central Asia as of the 2023 assessment.
This matters as context for the deployment picture. Kazakhstan is implementing AI in healthcare diagnostics — oncology centers receiving equipment that uses AI for reading X-rays, mammograms, CT, and MRI scans. Tajikistan’s zypl.ai is underwriting one quarter of all loans across eight financial institutions. Uzbekistan is deploying AI in tax administration and e-government. Each of these applications involves consequential decisions about people’s lives, and each of them is being deployed without any documented national standard for how the AI systems involved should handle bias, fairness, or discrimination. The GIRAI baseline records this not as a critique but as a starting point: the assessment makes visible what is not yet being asked, in a region where the asking has not yet started.
Other Thematic Areas: The Partial Picture
Several thematic areas show evidence in some countries and not others, in ways that reflect each country’s specific governance environment rather than a regional pattern.
Gender Equality returned positive evidence for Kyrgyzstan (government action and civil society) and not for any other country in the region. The government-side evidence reflects Kyrgyzstan’s national programmes on women’s economic participation in the digital economy, which include provisions on digital skills and access; civil society organizations working on digital inclusion and gender were also confirmed active in this space. The GIRAI assessment confirmed both actor categories.
Labour Protection returned civil society evidence for Kyrgyzstan — specifically through Tazabek, the Bishkek-based business and economics media outlet covering banking, real estate, energy, agriculture, and IT and telecommunications sectors. Tazabek publishes regular market pricing data updated multiple times daily and covers the effects of automated pricing systems on wages and market conditions — the kind of non-state, economics-focused reporting the GIRAI framework captures as civil society engagement on labour and AI. No Labour Protection evidence was found in any other country in the region.
Environmental Protection showed academic evidence in Kyrgyzstan — the country’s geography (over 90% of its territory above 1,500 metres, significant glacier coverage and high climate sensitivity) makes it a natural site for environmental AI research, and academic institutions confirmed active in this area at the time of the assessment. In Tajikistan, the AI strategy references green economy as a development goal and is cited in the EBRD Green City Action Plan for Dushanbe (2022) — a city-level planning document covering low-carbon transport, energy efficiency, green spaces, and climate resilience for the capital — which includes AI-adjacent smart city measures under the «Smart Regionalization» pillar. Neither country’s evidence constitutes a rights-protective framework; both represent the beginning of an evidence trail.
Health and Well-Being returned government action and academic evidence for Kyrgyzstan, grounded in the country’s digital health infrastructure work — telemedicine standardization and digital health records — and returned no evidence for any other country. This sits in an instructive contrast with Kazakhstan, which is implementing AI in healthcare diagnostics (oncology centers equipped with AI-enabled radiology systems), but where that deployment falls under the capacities picture rather than a rights-protective Health and Well-Being governance framework.
Children’s Rights and Indigenous Data Sovereignty returned zero evidence across the entire region — not because they are low-priority, but because neither has entered the documented AI governance conversation in any of the five countries.
What the Dimension Shows
The Human Rights and AI dimension is where the GIRAI framework is most effective at revealing what is not there. The five Central Asian profiles, read together, show a region where the AI rights conversation is either not happening, happening in limited domains (education, language), or happening in ways that have not yet produced the kind of documented, verifiable, actor-level evidence the index captures. Kyrgyzstan’s lead on this dimension is not because the country has solved the AI rights question; it is because it has more documented actors engaging with specific aspects of it. The regional pattern is a starting point, not a verdict.
Based on the GIRAI 1st Edition (2023) Human Rights and AI dimension findings for the five Central Asian states. Most granular data from the Kyrgyzstan GIRAI dashboard. Data source: global-index.ai. Regional hub: IDFI (Georgia). Kyrgyzstan country researcher: Aziz Soltobaev. This is a regional thematic comparison by KG Labs.
