Case-study focus: AI readiness, institutional capacity, data governance, public-sector implementation, responsible AI, local ecosystems, monitoring and evaluation.

Evidence base: The UNDP report draws on 26 Artificial Intelligence Landscape Assessments completed between 2024 and 2026. The countries are not presented as a statistically representative sample, ranked or assigned an aggregated readiness score.

Once AI adoption is underway, readiness becomes an implementation question rather than only a measure of preparation.

Pattern 1: Political attention is not the primary constraint

Political interest in AI is already visible in many assessed countries. The more difficult task is translating that interest into mandates, financing, delivery capacity, coordination, procurement capability, oversight and accountability.

Viet Nam: AI is being incorporated into strategies for research, development, digital transformation, socioeconomic development and national competitiveness. Constraints remain in regulation, funding, data, infrastructure and workforce capacity.

Malawi: High-level political recognition needs to be converted into strategy, delivery capacity, data governance, procurement reform and institutional coordination.

Ethiopia: Political commitment and state leadership are advancing AI as a national priority, while implementation depends on compute, interoperable data, applied skills and trust frameworks.

Trinidad and Tobago: Strong political commitment is accompanied by priorities relating to shared infrastructure, public-sector capability, trust, safety, talent retention and public awareness.

M&E implication: A strategy, public statement or national committee is not sufficient evidence of implementation readiness.

Pattern 2: AI adoption often enters through systems, not strategies

Consequential AI-related choices are frequently made through procurement, digital infrastructure, software and service-delivery systems before they are recognized as formal AI-policy decisions.

Uzbekistan: National AI strategies operate alongside sector priorities, e-government services, diagnostic pilots, virtual assistants and plans to expand AI-enabled public services.

Bhutan: AI adoption is expected to build on government-to-citizen services, payment gateways, government data infrastructure, national data exchange and digital identity.

Mongolia: AI technologies have entered some public services while institutional alignment, inter-agency arrangements and oversight remain works in progress.

Montenegro: AI adoption is connected to public-service integration, digital infrastructure, open data, public–private partnerships and potential sector applications.

M&E implication: Evaluators should examine procurement, registries, service portals, digital identity, payment systems, data exchanges and vendor relationships—not only national AI strategies.

Pattern 3: Data determines whether adoption becomes implementation

Strategies and pilot projects can create momentum, but implementation depends on data quality, accessibility, discoverability, interoperability, stewardship, security, privacy and responsible reuse.

Uzbekistan: Important digital-government assets coexist with fragmented ministerial data, heterogeneous formats, discovery gaps and inconsistent data-quality practices.

Burundi: Institutions produce increasing volumes of data, but access, quality, sharing, interoperability, security and protection require strengthening.

Côte d’Ivoire and Guinea: Priorities include data governance, shared platforms, institutional coordination, analytical capacity, cybersecurity and algorithmic responsibility.

Guatemala: Institutional silos, inconsistent quality, limited exchange standards and difficulty identifying available datasets constrain responsible adoption at scale.

M&E implication: Data readiness should be assessed as institutional implementation infrastructure, not only as a technical condition.

Pattern 4: Foundations determine agency, not just capacity

Foundations include connectivity, compute, cloud infrastructure, energy reliability, data centres, financing, technical talent, procurement expertise and institutional capability.

Bhutan: Strong connectivity and digital infrastructure are accompanied by a need for greater computing capacity and network resilience.

Malawi: Reliable electricity, connectivity, interoperable systems, data quality and trusted digital identity are important foundations for wider implementation.

Costa Rica: Favourable general digital conditions coexist with limitations in high-performance computing and AI-ready data-centre services.

Rwanda: Technical training and university programmes support adoption, while project management and ethical-governance capability are also required.

M&E implication: Infrastructure should be assessed according to whether it expands the ability to choose, adapt, supervise, negotiate, contest and sustain AI systems.

Pattern 5: Ecosystems determine whether foundations create local value

Infrastructure, data and talent generate development value when they are connected to public demand, research, financing, enterprise development, procurement opportunities and implementation pathways.

Montenegro: AI readiness is connected to innovation policy, public procurement, financing, competitiveness and public–private collaboration.

Malawi: Emerging talent and innovation nodes coexist with gaps in faculty capacity, compute access, financing, commercialization, public data access and startup procurement routes.

Dominican Republic and Colombia: Priorities include stronger financing, ecosystem coordination, entrepreneurship, private-sector innovation and links between research, training and productive-sector needs.

Paraguay: Wider participation, rights awareness and ethical-use training are emphasized across public institutions, universities, industry and civil society.

M&E implication: A successful initiative should leave local institutions and ecosystem actors better able to address future development problems.

Pattern 6: Governance becomes consequential through implementation

Responsible-AI principles become meaningful when they influence procurement, system selection, data use, human review, vendor management, monitoring, incident reporting, explanation, correction and redress.

Bhutan: Identified priorities include stronger oversight, legal frameworks, safety protocols, risk categorization, impact assessment, monitoring and explainability.

Mongolia: Gaps concern transparency, safety, accountability, explainability, risk classification and dedicated oversight.

Viet Nam: Relevant measures include transparency, bias mitigation, safety protocols, continuous evaluation and stakeholder engagement.

Trinidad and Tobago: Priorities include oversight, incident reporting, escalation, rights, recourse, explainability and opt-out mechanisms.

Dominican Republic: Responsible-AI intent is present, but accountability, human oversight, incident reporting and post-deployment monitoring remain weakly institutionalized.

M&E implication: Governance should be assessed by examining whether safeguards change contracts, procurement, monitoring, disclosure, performance standards and redress procedures.

The deeper risk is unmanaged adoption

AI may enter institutions through external providers, imported platforms or short-term projects without sufficient visibility into who benefits, who may be harmed, what dependencies are created, what capabilities are developed and who remains accountable.

Exercise 1: Identify what becomes binding

For each of the six readiness dimensions, identify:

1. Existing strengths
2. Binding constraints
3. Emerging dependencies
4. Potentially affected or excluded groups
5. Evidence still required
6. Priority institutional responses

Exercise 2: Compare two country examples

Select two countries from the case and compare:

• How AI is entering national systems
• Existing strategic or political commitment
• Relatively strong foundations
• Remaining data, infrastructure or institutional constraints
• Local participation and value creation
• Whether governance safeguards are operational
• Evidence required before adoption expands

Do not rank the countries. Examine how different combinations of strengths and constraints affect national agency.

Exercise 3: Build an AI-adoption evaluation matrix

Relevance: Is AI addressing a clearly defined development or public-service problem?

Coherence: Is the initiative connected to existing systems, strategies and institutional reforms?

Effectiveness: Is it improving the intended service or institutional outcome?

Efficiency: What savings and additional infrastructure, staffing, vendor and oversight costs are created?

Equity and inclusion: Who benefits, who may be excluded and which groups experience additional burdens?

Sustainability: Can national institutions supervise, maintain, adapt or replace the system?

Accountability: Can affected people obtain an explanation, challenge a decision and access correction or redress?

Final reflection

When an AI initiative is technically operational, but the responsible institution cannot independently evaluate it, supervise its provider, measure its distributional effects or correct its failures, should it be considered evidence of AI readiness?