Strategy for the Development of AI for the Period up to 2040<\/a>, adopted September 2022. It is also, as the GIRAI assessment makes clear, the country in the region where the gap between that strategy document and any actual governance framework is widest. The strategy is ambitious \u2014 5% of GDP from AI by 2040, four pillars covering smart cities, education, e-government, and industrial acceleration \u2014 but it contains no accountability provisions, no human oversight requirements, no safety standards, and stages the legal and institutional framework for later phases, after the technology is already deployed.<\/p>\n\n\n\nKazakhstan<\/strong> has the most developed policy pipeline: the 2018 Digital Kazakhstan program, a 2023 Concept on Digital Transformation and Cybersecurity, and an AI Concept for 2024\u20132029 that was under public discussion at the time of the assessment. The pace is real \u2014 but the GIRAI researcher’s assessment is equally direct: the recent instruments demonstrate accelerating ambition while containing a notable gap in ethical dimensions. The two-week public consultation window for the AI Concept is flagged as functionally narrow for a strategy document governing AI through 2029.<\/p>\n\n\n\nUzbekistan<\/strong> sits between the symbolic and the active. The 2021 Presidential Decree (PQ-4996) approved a two-year Programme of Measures covering eight directions \u2014 from drafting an AI development strategy to international cooperation \u2014 and the Advisory Council on Artificial Intelligence was established in August 2023, just before the GIRAI research window closed. What the instruments do not yet incorporate is the transparency, accountability, and public engagement layer that responsible AI governance requires. The framework outlines the creation of an AI ecosystem; it does not yet address the rights-protection architecture alongside it.<\/p>\n\n\n\nKyrgyzstan<\/strong> has no adopted AI legislation. The Digital Code \u2014 drafted in Chapter 23 with a risk-based AI governance framework (high-risk AI definitions, transparency requirements, human oversight, data quality standards) \u2014 was put to public consultation in August 2023 but had not been adopted as of February 2024. In a region where strategy documents sometimes appear without any enabling framework behind them, Kyrgyzstan’s position is reversed: the framework architecture is drafted and technically sophisticated, but the political consolidation has not arrived. The full GIRAI assessment shows this translated into the strongest Human Rights and AI dimension in the region \u2014 because the existing 100-plus fragmented legal acts do contain relevant provisions, even without a consolidated instrument.<\/p>\n\n\n\nTurkmenistan<\/strong> has nothing to assess in this dimension. No frameworks, no regulations, no strategies, no task forces, no public engagement. The GIRAI researcher described this as «near complete absence of evidence related to artificial intelligence.» The country’s information environment \u2014 documented internet shutdowns, government monopolization of telecommunications, no culture of government transparency \u2014 shapes not just the governance picture but the research itself. Absence of evidence is not evidence of absence; Turkmenistan’s profile is primarily a record of what responsible AI assessment looks like at the floor of what it can observe.<\/p>\n\n\n\nHuman Rights and AI: Where Rights Show Up<\/h2>\n\n\n\n
Kyrgyzstan’s Human Rights and AI dimension is the most detailed in the region, and the GIRAI assessment captures why: the combination of a functioning civil society, an academic sector engaged with the governance agenda, and national policy commitments that explicitly reference AI in relation to language and education creates a broader evidence base than any of its neighbors. Eight of twelve thematic areas returned positive evidence \u2014 Education (four actor categories), Cultural and Linguistic Diversity (three), Gender Equality (two), Health and Well-Being (two), Freedom of Expression (framework), Public Participation (government and academia), Labour Protection (civil society), and Environmental Protection (academia). Four areas were blank across all actors: Data Protection, Children’s Rights, Indigenous Data Sovereignty, and \u2014 most notably \u2014 Bias and Unfair Discrimination.<\/p>\n\n\n\n
Kazakhstan’s Human Rights and AI dimension is anchored in academia. The Ministry of Science and Higher Education identifies 24 universities and research centers active in AI research or development \u2014 a significantly deeper institutional footprint than any other country in the region. What the dimension does not show is non-state actor engagement meeting the index’s evidentiary standard: private sector and civil society activity on AI and human rights is constrained by Kazakhstan’s political environment. The documented use of facial recognition technology during the January 2022 civil unrest remains unresolved alongside the country’s institutional commitments.<\/p>\n\n\n\n
For Tajikistan, Uzbekistan, and Turkmenistan, the Human Rights and AI dimension is thin in ways that tell different stories. Tajikistan’s thinness reflects a governance conversation that treats critical examination of AI risks as an impediment to development \u2014 the researcher’s own description. Uzbekistan’s thinness reflects a policy landscape that has committed to AI deployment without yet incorporating human rights protections into the framework. Turkmenistan’s thinness is a function of information access: the country’s closed environment makes observation from open sources deeply limited, which is itself a finding.<\/p>\n\n\n\n
National AI Capacities: Institutions and Gaps<\/h2>\n\n\n\n
Kazakhstan’s lead on the National Responsible AI Capacities dimension is measurable: 24 universities and research centers engaged in AI work, four flagship institutions with advanced computing infrastructure (Eurasian National University, Al-Farabi Kazakh National University, Satpayev University, Nazarbayev University), and a demographic profile \u2014 more than 6.2 million people under 35, 39.3% with higher education degrees \u2014 that represents a workforce pipeline of genuine scale. The research and deployment work documented across these institutions spans several concrete domains. In healthcare<\/strong>, the Concept for Healthcare Development until 2026 commits to procurement of AI-enabled radiology, X-ray, and ultrasound equipment and software, with a focus on oncology diagnostics in remote regions. In e-government<\/strong>, the Ministry of Digital Development is piloting the automation of 1,414 state services using AI technologies. In manufacturing<\/strong>, AI applications for quality control and labour productivity are referenced in the national AI strategy pipeline. In statistical infrastructure<\/strong> \u2014 the unglamorous but foundational step for any governance evidence base \u2014 the National Statistics Committee added artificial intelligence, machine learning, and big data as recognized indicators in enterprise ICT surveys in 2022. The limitation is in the governance of that capacity: state investment is substantial, but participation from private sector and civil society in shaping its direction is constrained.<\/p>\n\n\n\nKyrgyzstan’s capacity story runs through language. The two binding national policy documents that explicitly name AI \u2014 the Cabinet Action Plan section on software development for Kyrgyz-language translation, and the State Language Policy Programme’s chapter on AI and natural language processing \u2014 gave Kyrgyzstan’s AI work a specific and defensible direction. This orientation has a longer history: KG Labs researcher Aziz Soltobaev had argued for the necessity of developing natural language processing for the Kyrgyz language in digital development strategy recommendations going back to 2018, identifying the absence of Kyrgyz-language AI tools as a structural gap in any digital inclusion agenda. The GIRAI assessment captures the policy documents that emerged from that direction \u2014 binding and in place, but ahead of the technical implementation they called for.<\/p>\n\n\n\n
Tajikistan’s most visible capacity contribution is commercial rather than academic: zypl.ai, the AI company founded by AI Council chair Azizjon Azimi, was providing AI underwriting for one quarter of all loans across eight financial institutions in Tajikistan at the time of the assessment. Telecom providers had built churn-prediction models in-house. The technology is ahead of the governance in Tajikistan in a way that is unusual even in a region where that gap is common. The country’s strategy targets 7 higher education institutions with AI departments by 2026 and 5,000 qualified AI specialists by 2040; neither target has a responsible AI qualification attached to it.<\/p>\n\n\n\n
Uzbekistan’s capacity is being built through deployment: the 2021 presidential programme identified pilot projects in finance, banking, tax administration, and e-government as the leading edge, driven by state priorities. The USAID Digital Ecosystem Country Assessment provides context for the underlying infrastructure. What the assessment found missing was civil society and academic participation in shaping the direction of that capacity \u2014 the ecosystem is being assembled from the top, and the feedback loops that would give it ethical content are not yet in place.<\/p>\n\n\n\n
Turkmenistan’s universities have some AI programme components. The GIRAI assessment confirmed this and also confirmed that none of them incorporate AI ethics or responsible use. There is no academic publishing on AI, no expert participation in regional or international forums, no advocacy on AI governance \u2014 the academic component exists as an island without the connective tissue that would make it part of a national AI ecosystem.<\/p>\n\n\n\n
What the Regional Pattern Shows<\/h2>\n\n\n\n
Read together, the GIRAI 2023 Central Asian profiles show a region at different moments in the same transition \u2014 from digital infrastructure investment toward something that could be called responsible AI governance. No country in the region had completed that transition by the time of the assessment. The question the baseline captures is not whether any country has arrived, but where each one was standing when the first measurement was taken, and what the distance to a rights-protective framework looked like from there. The five country profiles in this series \u2014 Kyrgyzstan<\/a>, Kazakhstan<\/a>, Tajikistan<\/a>, Uzbekistan<\/a>, Turkmenistan<\/a> \u2014 are each a starting point. Subsequent GIRAI editions will be able to measure movement from it.<\/p>\n\n\n\n
\n\n\n\nBased on the GIRAI 1st Edition (2023) country profiles for all five Central Asian states. Data source: global-index.ai<\/a>. Regional hub: IDFI (Georgia). Kyrgyzstan country researcher: Aziz Soltobaev. Other country profiles read as observational coverage. This is a regional comparison post by KG Labs as part of its Central Asia AI governance series.<\/p>\n\n","protected":false},"excerpt":{"rendered":"Central Asia in the GIRAI 2023 Assessment: What the Regional Picture Shows The Global Index on Responsible AI (GIRAI) 1st Edition assessed 138 countries across three dimensions \u2014 Responsible AI Governance, Human Rights and AI, and National Responsible AI Capacities. All five Central Asian states were included: Kyrgyzstan, Kazakhstan, Tajikistan, Uzbekistan, and Turkmenistan. KG Labs […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"inline_featured_image":false,"footnotes":""},"categories":[246,375,357],"tags":[487,444,484,644,493,13,516,637,673,485,504,514,509],"class_list":["post-7792","post","type-post","status-publish","format-standard","hentry","category-ai-governance","category-policy-and-regulation","category-research-and-evidence","tag-ai-policy","tag-central-asia","tag-girai","tag-series-girai-2023","tag-kazakhstan","tag-kyrgyzstan","tag-regional-comparison","tag-format-research","tag-op-research-evidence","tag-responsible-ai","tag-tajikistan","tag-turkmenistan","tag-uzbekistan"],"translation":{"provider":"WPGlobus","version":"3.0.2","language":"ru","enabled_languages":["en","ru"],"languages":{"en":{"title":true,"content":true,"excerpt":false},"ru":{"title":false,"content":false,"excerpt":false}}},"_links":{"self":[{"href":"https:\/\/kglabs.org\/ru\/wp-json\/wp\/v2\/posts\/7792","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/kglabs.org\/ru\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/kglabs.org\/ru\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/kglabs.org\/ru\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/kglabs.org\/ru\/wp-json\/wp\/v2\/comments?post=7792"}],"version-history":[{"count":1,"href":"https:\/\/kglabs.org\/ru\/wp-json\/wp\/v2\/posts\/7792\/revisions"}],"predecessor-version":[{"id":7835,"href":"https:\/\/kglabs.org\/ru\/wp-json\/wp\/v2\/posts\/7792\/revisions\/7835"}],"wp:attachment":[{"href":"https:\/\/kglabs.org\/ru\/wp-json\/wp\/v2\/media?parent=7792"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/kglabs.org\/ru\/wp-json\/wp\/v2\/categories?post=7792"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/kglabs.org\/ru\/wp-json\/wp\/v2\/tags?post=7792"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}