2025年数字发展与趋势报告_夯实人工智能基础_135页_3mb

Summary of Digital Progress and Trends Report 2025: Strengthening AI Foundations

Core Content

The Digital Progress and Trends Report 2025: Strengthening AI Foundations explores the role of artificial intelligence (AI) in global development and outlines the key challenges and opportunities for low- and middle-income countries (LICs and MICs) to harness AI effectively. The report emphasizes the need for strategic investment in four foundational elements—Connectivity, Compute, Context, and Competency—to ensure equitable and sustainable AI adoption.

Main Messages

• AI presents both opportunities and risks, including inequality, ethical concerns, and the concentration of innovation in high-income countries (HICs).
• Small AI offers a practical and accessible way for LICs and MICs to begin leveraging AI for development with limited resources.
• AI adoption is uneven, with HICs leading in innovation, research, and infrastructure, while LICs and MICs lag in access and capacity.
• Four foundational investments (4Cs) are critical to enable AI adoption and ensure that AI benefits all countries, not just a few.

Key Trends and Findings

1. AI Innovation and Adoption

• AI innovation is highly concentrated in HICs, with:
  • 87% of notable AI models developed in HICs.
  • 86% of AI start-ups located in HICs.
  • 91% of venture capital (VC) funding in AI start-ups sourced from HICs.
• AI-related job postings are growing faster in MICs than in HICs:
  • 16% growth in upper-middle-income (UMIC) countries.
  • 11% growth in lower-middle-income (LMIC) countries.
  • Only 2% growth in HICs.
• LICs remain at the periphery of AI innovation, with less than 1% of global AI traffic and negligible participation in AI start-ups and funding.

2. Connectivity

• Internet access is expanding globally, but disparities remain:
  • HICs: 93% internet penetration (2024).
  • UMICs: 81%.
  • LMICs: 54%.
  • LICs: 27%.
• Data consumption is also uneven:
  • HICs: 1,400 GB per capita (2023).
  • LMICs: 100 GB.
  • LICs: 5 GB.
• Satellite technology is helping to close the digital divide, with 5,300 commercial communication satellites in orbit as of 2023.

3. Compute

• Compute power is essential for AI, but it is concentrated in HICs:
  • 50% of secure internet servers are in the U.S.
  • 41% in other HICs.
  • 9% in the rest of the world.
• Cloud computing exports are dominated by the U.S.:
  • 87% of global cloud computing exports in 2023.
  • 84% of these exports go to HICs.
  • Only 16% reach MICs, and negligible to LICs.
• Domestic compute capacity is a strategic decision for developing countries, as they face challenges in accessing affordable international cloud services.

4. Context

• AI model effectiveness depends on training data:
  • 85% of AI training data start-ups are in HICs.
  • English dominates open-source data sets (over 50%).
• LICs and MICs are exploring localized data formats such as video and audio to increase participation.
• Adapting global models to local contexts is crucial for developing effective, inclusive AI systems.

5. Competency

• Digital skills are becoming a baseline requirement across many occupations.
• Skill gaps are significant in LICs and MICs:
  • Basic digital skills: 66% in HICs, <5% in LICs.
  • Advanced skills: <10% in HICs, <1% in LICs.
• AI job vacancies are rising faster in MICs than in HICs, with:
  • 16% growth in UMICs.
  • 11% growth in LMICs.
  • 2% growth in HICs.
• Brain drain further hinders AI development, with talent outflows 3–4 times higher than inflows in some countries.

Strategic Recommendations

• Invest in the 4Cs to build a foundation for AI adoption.
• Fiscal policies should reduce inequality and provide safety nets amid AI disruptions.
• Labor and education policies must focus on future-proofing workers and improving digital education.
• Trustworthy and responsible AI systems are essential to manage risks and ensure ethical use.
• Open-source technologies can help developing countries adapt AI to local needs without reinventing the wheel.

Conclusion

The report underscores the urgency of strategic AI investment, particularly for LICs and MICs, to ensure inclusive and sustainable development. While AI has the potential to drive economic transformation and create new opportunities, its benefits can only be realized through equitable access, localized adaptation, and skill development. The World Bank Group is committed to supporting these efforts and promoting AI as a tool for shared prosperity.