Methodology, Conceptual Framework, Advantages, and Recommendations for Uzbekistan’s Inclusion in Future Editions
The Artificial Intelligence and Democratic Values Index (AIDVI) is a global index published annually since 2020 by the Center for AI and Digital Policy (CAIDP). It evaluates the AI policies and practices of countries around the world based on their alignment with human rights and democratic values.
In the latest edition of the index, published in 2026, 90 countries were evaluated. Uzbekistan has not yet been included in the index.
Key Indicators and Standards
- Core Focus: AIDVI focuses on human rights, the rule of law, and democratic governance indicators.
- Primary Criteria: Key indicators include supporting and implementing the OECD/G20 AI Principles and UNESCO's Recommendations on the Ethics of AI.
- Public and Institutional Measures: Other indicators include public participation opportunities in shaping national AI policies, the adoption of algorithmic transparency laws, and the establishment of independent agencies to address AI-related issues.
- Economic Exclusions: Although patents, publications, investments, and impacts on employment are crucial indicators for the AI economy, they are not considered in this index.
The AIDVI bases its evaluation framework on the standards of the following international documents and organizations:
- OECD/G20 AI Principles (2019, updated in 2024): The first set of intergovernmental standards for Artificial Intelligence. It requires AI systems to be human-centric, transparent, accountable, safe, and inclusive.
- Universal Declaration of Human Rights (UDHR): Considered the fundamental international framework for human rights. The index uses this document as a benchmark (primary criterion) when evaluating countries.
- UNESCO Recommendation on the Ethics of AI (2021): A comprehensive global framework requiring compliance with ethical rules throughout the entire lifecycle of AI.
- Council of Europe AI Treaty (CETS No. 225, 2024, entering into force: 2025): The first international, legally binding AI treaty guaranteeing human rights, democracy, and the rule of law. By the beginning of 2026, 45 countries had signed this document.
- European Union AI Act (2024, implementation: 2025–2027): The world's first major legislative framework regulating AI technologies based on risk levels.
- Global Privacy Assembly (GPA) Resolutions: A series of critical resolutions aimed at ensuring privacy and accountability in AI: 2018 (AI and Ethics), 2020 (AI and Accountability), 2022 (Facial Recognition Technology), and 2023 (Generative AI).
- Freedom House: CAIDP relies on this organization's country ratings ("Free", "Partly Free", "Not Free") to assess the state of human rights.
Evaluation Methodology
The AIDVI consists of 12 evaluation criteria. Countries are evaluated on a three-tier system for each criterion. All 12 criteria carry equal weight, with a maximum total score of 12 points.
- Y (Yes, fully) – 1.0 point
- P (Partially) – 0.5 point
- N (No) – 0.0 point
Table 1. Full Description of the 12 Evaluation Criteria for CAIDP AIDVI 2026
| No | Criterion Name | Essence of Evaluation and Sources | Score Type |
|---|---|---|---|
| Q1 | Support for OECD/G20 AI Principles | Is the country an OECD member, an adhering country to OECD principles, or a member of G20/GPAI? Note: Primarily a binary evaluation. | Y or N |
| Q2 | Implementation of OECD Principles | Is the country applying these principles in practice? Determined based on OECD reports, government documents, and NGO sources. | Y/P/N |
| Q3 | Support for the Universal Declaration of Human Rights | Has the country signed or recognized the UDHR? Binary evaluation. (Exceptions like Saudi Arabia are reviewed separately). | Y or N |
| Q4 | Implementation of UDHR (State of Human Rights) | Based on Freedom House ratings: • "Free" = Y • "Partly Free" = P • "Not Free" = N |
Y/P/N |
| Q5 | Public Participation in Developing National AI Policy | Were official consultation processes conducted? Was real and meaningful public participation ensured? Are public opinions reflected in the national AI strategy? | Y/P/N |
| Q6 | Public Openness of AI Policy Documents | Are government AI policy documents complete, clear, and openly posted on state websites? Is an official English translation available? (Absence makes independent evaluation difficult). | Y/P/N |
| Q7 | Existence of an Independent Body for AI Oversight | Is there an independent and influential oversight mechanism, such as a data protection agency, a human rights commission, or an AI ethics commission? | Y/P/N |
| Q8 | Key Democratic Objectives in National AI Policy | Are the five core objectives clearly outlined in the national AI strategy: Fairness, Accountability, Transparency, Rule of Law, and Fundamental Rights? | Y/P/N |
| Q9 | Legal Guarantee of the Right to Algorithmic Transparency | Does the EU GDPR (Article 22), Council of Europe Convention 108+ (Article 9), or national legislation establish the right to algorithmic transparency? (EU members and 108+ signatories automatically receive Y). | Y/P/N |
| Q10 | Support for the Council of Europe AI Treaty (CETS No. 225) | This criterion replaced the UNESCO recommendation. Countries that signed the treaty = Y; EU member states (via EU signature) receive = P. (45 countries signed by early 2026). | Y/P/N |
| Q11 | Implementation of UNESCO's Recommendation on AI Ethics | To receive a Y, the RAM (Readiness Assessment Methodology) must be completed along with additional political steps. To receive a P, the RAM process must be initiated or aligned actions must be taken. (Relies on UNESCO monitoring and government reports). | Y/P/N |
| Q12 | Compliance with Global Privacy Assembly (GPA) Resolutions | Has the country's data protection authority supported the 4 GPA resolutions (2018 - Ethics, 2020 - Accountability, 2022 - Facial Recognition, 2023 - Generative AI)? • All four = Y • 1 to 3 = P • None = N |
Y/P/N |
Tier System
After being evaluated according to the criteria, countries are placed into tiers based on their total score:
Table 2. AIDVI Tier System
| Tier | Score Range | Description |
|---|---|---|
| Tier I | 10.5–12.0 | Best Practice: A strong, formally documented, and fully implemented approach in practice across all key aspects of AI policy. |
| Tier II | 9.0–10.0 | Strong Approach: The overall situation is positive, but certain shortcomings exist in some areas. |
| Tier III | 7.0–8.5 | Evolving: There are clear achievements in AI policy, but systemic gaps remain significant. |
| Tier IV | 5.0–6.5 | Limited Approach: Many key elements are missing; AI policy is still in its formative stages. |
| Tier V | 0.0–4.5 | Underdeveloped: Severe systemic shortcomings exist in AI governance and democratic safeguards. |
Advantages and Disadvantages of AIDVI
- Advantages: The primary benefits acknowledged are its complete transparency (the 12 criteria and scoring system are public, allowing states to independently verify their ratings) and its two-layered de jure and de facto approach (it checks not only the existence of laws, but also their actual implementation via NGO and independent media sources).
- Disadvantages: Weaknesses include its low global coverage (covering less than half of the world's countries; Uzbekistan and many Central Asian states are excluded due to a lack of local experts) and its reliance on Western-centric models (criteria like GDPR and Freedom House risk undervaluing real achievements in other regions' specific cultural and political contexts).
Current Regional Context: In the 2026 AIDVI report, Kazakhstan is the only Central Asian country included. It scored 4.5 out of 12 points, placing it in the lowest group—Tier V (Countries with concerning situations).
Given that Uzbekistan and Kazakhstan score closely in most international indexes, Uzbekistan would likely land in the lower tiers (Tier IV or Tier V) if it were added to the ranking today without changes.
Practical Recommendations for Including Uzbekistan
- Establish an Expert Corps: Establish a system to send specialists from the Ministry of Digital Technologies, its subordinate organizations, or leading higher education institutions to the annual CAIDP Policy Clinic program. This will build a local team of experts meeting international standards, whose independent reports can be integrated directly into the index database, mitigating regional expert shortages.
- Complete UNESCO's RAM: Uzbekistan should become one of the first in the region to finalize UNESCO's Readiness Assessment Methodology (RAM) process. Beyond securing a high score for criterion Q11, this would demonstrate the republic's regional leadership in digital technologies on an international scale.
- Sign the Council of Europe AI Treaty: Study and explore possibilities to sign the Council of Europe AI Treaty (CETS No. 225). This treaty is open to non-member states, and signing it would grant Uzbekistan a direct, maximum 1.0 point (Y) for criterion Q10.
- Ensure De Facto Implementation and Openness: Since the index measures real-world enforcement, civil society institutions and independent experts must be officially included in National AI councils and working groups. Public consultations and their final results should be regularly published on state resources in English, boosting performance in criteria Q5 and Q6.
Comprehensive implementation of these recommendations could lay the groundwork for Uzbekistan to enter at least Tier IV when it is integrated into the index.
References
-
Resolution of the President of the Republic of Uzbekistan No. PP-358 “On Approval of the Strategy for the Development of Artificial Intelligence Technologies until 2030.” [Link: lex.uz]
-
CAIDP (2026). Artificial Intelligence and Democratic Values – CAIDP AI Index 2026. Center for AI and Digital Policy, 2026. [Link: caidp.org]
-
CAIDP (2025). AI and Democratic Values Index 2025. Center for AI and Digital Policy, 2025. [Link: caidp.org]
-
OECD AI Observatory. “OECD AI Observatory Index: A comprehensive view of national AI ecosystems”. [Link: oecd.org]
-
UNESCO. “Readiness Assessment Methodology (RAM): Ethics of Artificial Intelligence”. [Link: unesco.org]
-
Freedom House. Freedom in the World (Annual Reports). [Link: freedomhouse.org]
-
Council of Europe (2024). Framework Convention on Artificial Intelligence and Human Rights, Democracy and Rule of Law (CETS No. 225). [Link: coe.int]
-
CAIDP. Universal Guidelines for Artificial Intelligence (UGAI). [Link: caidp.org]
-
Global Privacy Assembly (GPA). Resolutions on AI (2018, 2020, 2022, 2023). [Link: globalprivacyassembly.org]
