Suggested Sierra Leone National Artificial Intelligence Policy (High-Level Draft)

I did another weekend brainstorming and used AI to help with a high-level artificial intelligence draft policy for Sierra Leone.

Vision

To use practical, affordable and secure artificial intelligence to improve education, healthcare, agriculture, public administration and economic productivity, whilst ensuring national data sovereignty.

Guiding Principle

“Small Models, Local Infrastructure, Real Problems.”

Sierra Leone should prioritise specialised AI systems that can operate on local infrastructure rather than relying exclusively on large foreign cloud services.

Strategic Priorities

Local AI Infrastructure

• Establish a National AI Computing Centre hosted within Sierra Leone. • Prioritise on-premises and private cloud deployments for government systems. • Encourage the use of energy-efficient AI systems that can operate on modest hardware.

Open-Source Foundation Models

Government agencies shall prioritise the evaluation of open-source AI models before procuring proprietary alternatives.

Initial recommended models include:

• DeepSeek — for reasoning, coding, administration and educational applications. • Qwen — for multilingual assistants, knowledge systems and local fine-tuning.

These models should be adapted to Sierra Leonean use cases rather than deployed as general-purpose chatbots.

National AI Applications

Priority sectors:

Education • Exam tutoring assistants. • Teacher support systems.

Healthcare • Clinical decision-support tools. • Medical knowledge assistants.

Agriculture • Crop advisory services. • Pest and disease identification.

Government • Citizen service assistants. • Document search and summarisation.

Sierra Leone Knowledge Models

Develop specialised national AI models trained on:

• School curricula. • Government publications. • Agricultural guidance. • Health information. • Laws and regulations.

The objective is to achieve accuracy in local knowledge rather than broad general intelligence.

Data Sovereignty and Security

• Sensitive government data should remain within Sierra Leone-controlled infrastructure wherever practicable. • Citizens’ personal information must not be used to train public AI systems without appropriate authorisation. • Government AI deployments must maintain audit logs and human oversight.

Skills Development

• Introduce practical AI literacy programmes in schools and universities. • Support training for teachers, civil servants and software developers. • Encourage local innovation through the use of open-source AI technologies.

Implementation Approach

Phase 1 (Years 1–2) • Establish the National AI Computing Centre. • Pilot education and government AI applications. • Fine-tune DeepSeek and Qwen for local use.

Phase 2 (Years 3–5) • Expand sector-specific AI assistants. • Develop national knowledge models. • Support local start-ups building AI applications.

Success Measure

The success of this policy shall be measured not by the size of AI models, but by improvements in educational outcomes, healthcare delivery, agricultural productivity, government efficiency and job creation.

Wilfred is not as clever as he pretends to be.