A Blueprint for Intelligent Economies 2024

Spotlight on three strategic objectives This section explores the three strategic objectives commonly prioritized in all geographic regions.2 While all nine strategic objectives must be considered holistically to drive an effective AI strategy, feedback from government, enterprises and academic stakeholders involved in the development of this white paper identified three that are the most often prioritized within national AI programmes: building sustainable AI infrastructure, curating diverse and high-quality datasets, and establishing guardrails for ethics, safety and security. This section provides a comprehensive analysis of these three objectives, including an outline of the key challenges, necessary enablers and examples of practical decisions that stakeholders can make. 2.1 Build sustainable AI infrastructure Delivering sustainable and resilient AI infrastructure will require significant investment and cross-sector collaboration to create scalable, secure and environmentally responsible systems. The challenges are complex and multifaceted, requiring new capabilities involving the coordination of efforts at global, regional and national levels. Key challenges in building sustainable AI infrastructure TABLE 1 Key challenges Examples of successful initiatives High energy consumption and environmental impactAdvanced energy programmes to power data centres: Collaborative agreements between data centre firms and energy providers can support the deployment of sustainable energy such as wind, solar or nuclear to power the massive demands for compute. Energy efficient AI model development: Organizations are complementing large models with smaller sector-specific models trained on narrow datasets to optimize energy consumption during large language model (LLM) training. AI optimized energy consumption: Use AI to optimize energy management by predicting consumption patterns, forecasting demand requirements and automating distribution. Significant scale of investments requiredRegional sharing of AI infrastructure: Data centre capacity is being shared between countries via regional clusters and arrays. Incentives for private sector investment: Governments are introducing tax incentives and financial grants and creating the wider enabling environment (through AI national strategies, regulatory reforms, capacity building, etc). Non-secure and non-resilient AI supply chainsInternational trade corridors: Trade agreements between international partners, bilateral or multilateral agreements, and lists of trusted suppliers are enabling resilient and diverse AI supply chains. National AI clusters: Governments are facilitating the development of clusters that enable the research, design and manufacturing of AI critical hardware to stay onshore. AI cloud resources: Collaboration with home-grown technology firms and global leaders of cloud infrastructure are contributing to creating trusted national sovereign clouds. A growing digital gapPartnerships with network providers: Some markets are now providing access to emerging high-speed and full coverage internet networks such as satellite internet constellation or collaborating with telecommunication providers for mobile internet deployment. High cost of current generation of internet devicesDevice subsidy programmes: Subsidized low-cost devices and network connectivity are supporting low-income and digitally disadvantaged groups. Low-cost AI optimized devices: Collaboration with technology providers and non-profit organizations is providing widespread access to devices for impactful AI use cases. Blueprint for Intelligent Economies 8