Research Summary: RADAR: Readiness for AI Discovery and Agentic Reach

A new research framework, RADAR (Readiness for AI Discovery and Agentic Reach), has been introduced to assess how effectively artificial intelligence systems can interact with public services across 166 countries. It evaluates both the ability of AI chatbots to provide correct, officially sourced information about public services (informational legibility) and the capability of automated agents to access and act on these services (agent operability). The primary finding is a significant disparity: AI systems are considerably better at describing public services than at actually interacting with them, a gap that persists regardless of national wealth.

Why it matters

This research provides a critical benchmark for governments and international bodies evaluating the practical implementation of AI in public service delivery. Understanding the significant gap between AI's informational capacity and its operational reach is crucial for strategic planning in digital transformation initiatives and for setting realistic expectations for AI's current capabilities in citizen engagement.

Key insights

• The RADAR framework measures AI system effectiveness in government-citizen interactions across 166 countries.
• It assesses two key areas: 'informational legibility' (AI chatbot's ability to provide correct, country-specific public service information) and 'agent operability' (automated agent's ability to reach and act on a service).
• A central finding is that AI can describe public services much better than it can act upon them.
• In all 166 countries surveyed, informational legibility scores consistently surpass agent operability scores.
• This gap between informational legibility and agent operability does not diminish with increased national wealth.
• Factors such as national income and language explain only a partial amount of the observed differences.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2609.28480