G7 toolkit for artificial intelligence in the public sector; report prepared for the 2024 Italian G7 presidency and the G7 digital and tech working group

Related Organizations

Table of Contents

• G7 Toolkit for AI in the Public Sector 3
• Preface 4
• Executive summary 7
• Key messages 8
• Establish clear strategic objectives and action plans in line with expected benefits 8
• Include the voices of users in shaping strategies and implementation 8
• Overcome siloed structures in government for effective governance 8
• Establish robust frameworks for the responsible use of AI 8
• Improve scalability and replicability of successful AI initiatives 8
• Enable a more systematic use of AI in and by the public sector 9
• Adopt an incremental and experimental approach to the deployment and use of AI in and by the public sector 9
• 1 Introduction and background 10
• 2 Enabling safe, secure, and trustworthy AI systems in the public sector 12
• 2.1. National strategies and policies for AI in the public sector 13
• 2.1.1. Key objectives and actions covered by AI strategies 15
• 3.1.1.1 Talent and skills 16
• 3.1.1.2 Procurement and partnerships 17
• 3.1.1.3 Guidance for AI development deployment and use 18
• 3.1.1.4 Government data in AI applications 20
• 3.1.1.5 Supporting infrastructure 21
• 3.1.1.6 General government functions 22
• 2.1.2. The role of public consultations and stakeholder engagement in national strategies 22
• 2.2. Governance frameworks: institutional arrangements and coordination mechanisms 24
• 2.2.1. Multi institutional governance approach 25
• 2.2.2. Single lead institutional governance approach 26
• 2.3. Safeguards and guardrails 27
• 2.3.1. Promoting transparency in public algorithms 29
• 2.3.2. Guidance on the use of AI in and by the public sector 30
• 3 Current trends in AI in the public sector 34
• 3.1. Expected benefits and impacts 34
• 3.1.1. Efficiency of public sector internal operations 36
• 3.1.2. Responsiveness of public service delivery 37
• 3.1.3. Improving accountability in government. 38
• 3.1.4. Effectiveness of policymaking 40
• 3.2. Policy options to address key implementation challenges 43
• 3.2.1. Challenge 1. Strengthening infrastructure 43
• 3.2.1.1 Policy option 1. Data storage solutions 44
• 3.2.1.2 Policy option 2. Data sharing solutions and frameworks 45
• 3.2.1.3 Policy option 3. Testing, experimentation, and support infrastructures 48
• 3.2.2. Challenge 2. Procuring AI and partnering outside the public sector 49
• 3.2.2.1 Policy option 1. Tools and requirements for public procurement of AI 50
• 3.2.2.2 Policy option 2. Public-private partnerships 53
• 3.2.3. Challenge 3. Upskilling the public sector 54
• 3.2.3.1 Policy option 1. Sharing of best practices 56
• 3.2.3.2 Policy option 2. Training and upskilling initiatives 57
• 3.2.3.3 Policy option 3. AI Competencies frameworks 59
• 3.2.3.4 Policy option 4. Hiring and retaining AI talent 61
• 3.2.4. Challenge 4. Establishing frameworks for data governance in the public sector 62
• 3.2.4.1 Policy option 1. Government data strategies 63
• 3.2.4.2 Policy option 2. Data leadership 65
• 3.2.4.3 Policy option 3. Data management and quality frameworks 66
• 3.2.4.4 Policy option 4. Privacy and personal data protection frameworks 68
• 3.2.5. Challenge 5. Monitoring AI implementation in the public sector 70
• 3.2.5.1. Policy option 1. Monitoring AI investments 70
• 3.2.5.2. Policy option 2. Oversight and monitoring of AI systems 74
• 4 Mapping the journey for AI solutions in the public sector 76
• 4.1. Framing 77
• 4.2. Ideating 78
• 4.3. Prototyping 78
• 4.4. Piloting 79
• 4.5. Scaling up 79
• 4.6. Monitoring (cross-cutting action) 80
• 4.6.1. Quality and performance metrics 80
• 4.6.2. AI assurance 80
• 4.7. Engaging (cross-cutting action) 81
• References 82
• Endnotes 90