Summary of Key Points
A report jointly produced by the World Economic Forum (WEF) and Capgemini aims to assist governments in transitioning from occasional use of AI to large-scale adoption for value creation. The report argues that previous digital initiatives merely moved paper-based processes onto screens, whereas “agentive AI” – intelligent assistants capable of planning and working across departments – represents a fundamental shift. Such AI not only analyzes data but can also provide end-to-end services (e.g., automatically issuing certificates). The report outlines four areas for government preparation and assesses the suitability of 70 government functions for AI integration, categorizing them as high/medium/low readiness. It cites successful cases from Ukraine, Germany, and the UAE, predicting that the use of AI in public sectors could generate $9.8 trillion in value by 2034.
Agentive AI: More Than Just a Screen
Traditional digital transformations, such as replacing paper forms with online ones, still require individuals to carry out the processes themselves. Agentive AI, on the other hand, acts like a “super assistant”:
- Cross-departmental coordination: For example, when requesting an income certificate, AI can automatically retrieve data from tax and social security agencies, eliminating the need for separate visits.
- Autonomous decision-making: In Germany, building permit-related AI systems review applications against legal regulations and provide preliminary approval opinions.
- Proactive service: Diia.AI in Ukraine reminds users of upcoming social security renewals and even handles the renewals on their behalf.
In other words, while previously citizens had to approach the government for services, with agentive AI, the government is now taking the initiative to assist them.
Four Steps Governments Need to Take
To adapt to agentive AI, governments must break with old habits:
1. Shift from departmental silos to functional collaboration: Data from multiple departments (e.g., police, civil affairs, social security) must be shared to enable AI to work across them.
2. Clarify priorities and sequences: AI should be implemented in areas where it can clearly save money and time (e.g., handling inquiries, issuing certificates), with clear accountability for decisions.
3. Move from occasional trials to systematic deployment: Instead of testing AI in a single department, integrate it into core workflows to create an “intelligent nation.”
4. Be cautious and steady: Although AI technology develops rapidly, governments must consider institutional factors (e.g., data privacy) and staff capabilities to avoid risks.
Which Government Functions Are Suitable for AI?
The report categorizes 70 government functions into three groups based on potential and complexity:
- High readiness (10 functions, 14%): High potential and low complexity, such as issuing certificates and handling inquiries. These tasks have clear rules and are frequently performed, making them ideal for AI implementation.
- Medium readiness (25 functions, 36%): Require certain conditions, such as improving data quality or governance rules before adopting AI.
- Low readiness (35 functions, 50%): High complexity and uncertain potential, such as policy planning and security responses. These tasks currently require human judgment and can be piloted gradually.
Global Examples
These cases demonstrate how AI can significantly save time and money for governments while enhancing citizen convenience:
- Diia.AI (Ukraine): A national-level AI assistant that automatically issues income certificates and handles inquiries. Users provide natural language commands, and AI generates official documents within minutes; it has processed over 1 million inquiries, issuing more than 7,000 certificates.
- German building permit AI: Speeds up the approval process by automatically reviewing applications against legal requirements, reducing the need for extensive data analysis.
- UAE HR AI: Helps government employees with 130 HR-related tasks, freeing up time for talent management.
These examples show that AI can genuinely improve efficiency and citizen experience.
Conclusion
AI presents significant opportunities for governments, but progress must be gradual. Fifty percent of core government functions are suitable for AI integration, and 90% of public sectors plan to adopt agentive AI within two to three years, with potential benefits reaching $9.8 trillion by 2034. The key is for governments to prepare properly (by breaking down silos and establishing clear strategies) and start with simpler tasks before moving on to more complex ones. After all, AI is a tool that must be used effectively to truly benefit the public.