The Promise of Artificial Intelligence

Governments should prioritize investments in foundational infrastructure and human capital.

• AI adoption depends on strong enabling conditions. 
• Affordable electricity, reliable internet connectivity, digital devices, and foundational skills remain essential prerequisites for businesses and workers to benefit from AI. 
• Without these foundations, the gains from AI will be limited and unevenly distributed. 

Beyond adoption, countries need policies that help domestic firms customize AI solutions for local needs and languages. This requires: 

• Investments in data ecosystems, compute infrastructure, advanced skills, entrepreneurship, and competitive markets. 
• Improved access to finance, and policies that connect AI solution providers with potential users.
• Supporting innovation through procurement reforms.
• Stronger university-industry collaboration.

Governments generate the greatest value from AI when they deploy it to solve defined public-sector challenges. Successful implementation requires: 

• Matching AI ambitions with institutional readiness, data availability, budget capacity, and governance arrangements. 
• Building AI-ready data systems: digitizing admin records, strengthening management information systems, improving data quality, and enabling safe data sharing across agencies
• Strengthening digital public infrastructure.
• Investing in staff capabilities and establishing sustainable operating models for AI systems rather than treating AI as a one-time technology purchase.

AI is only as effective as the systems that support it. To maximize effectiveness, governments should:

• Establish mechanisms to evaluate AI performance, verify outcomes, and monitor impacts to help identify what works, correct failures, and improve public trust.
• Use government procurement to crowd in local AI solutions in high-stakes applications, such as education and health, where local contexts matter most. 

Trust is the foundation of successful AI adoption. Governments should take a precautionary approach to AI governance by:

• Introducing safeguards before harms become widespread. 
• Assessing governance gaps, clarifying responsibilities, and strengthening enforcement capacity.
• Developing governance frameworks that evolve as technology changes. 

Voluntary standards are an important starting point to support trustworthy AI, but they should be complemented by effective public oversight and accountability mechanisms.

International cooperation can help countries:

• Avoid fragmented regulatory approaches.
• Strengthen enforcement.
• Prevent a race to the bottom on safeguards.
• Ensure developing countries have a voice in shaping global AI rules.