Jordan launched its first e-government portal in the early 2000s. Two decades later, the ambition is significantly higher: the government has committed to making 80% of public services available digitally, and the Sanad platform has brought hundreds of services under a single interface. The next phase isn't just about moving services online — it's about making them intelligent.
Where AI Creates Real Value in Government Services
The highest-value AI applications in citizen services aren't the futuristic ones — they're the ones that remove the most friction from the most common interactions. Three categories consistently deliver measurable impact:
- Document processing and classification. Government agencies receive enormous volumes of documents — permit applications, license renewals, court filings, social service requests. Most of this intake is still manual. AI-powered document intelligence can extract structured data from unstructured forms, classify requests by type and urgency, and route them to the correct department without human touch. The time savings are immediate and significant.
- Intelligent query handling. Citizens calling government contact centers typically have repetitive, predictable questions: what documents do I need, what's the status of my application, how do I appeal a decision. AI assistants trained on accurate, up-to-date government data can handle the majority of these queries in Arabic, freeing agents for genuinely complex cases.
- Proactive eligibility and notification. Rather than requiring citizens to discover services they qualify for, AI can identify eligible individuals from existing government databases and proactively notify them. This approach has been used effectively in social protection programs to reach vulnerable populations who weren't aware of their entitlements.
The Constraints That Matter
Building AI for the public sector has constraints that don't apply in commercial settings. Explainability is one: when an AI system makes a determination that affects a citizen's access to a service, that decision must be explainable in plain language and auditable after the fact. A model that performs well but can't explain its reasoning is not deployable in a government context.
Bias and equity are another constraint. AI systems trained on historical government data will reflect historical patterns — which may include underservice of certain communities. Before deployment, any model used in citizen-facing decisions needs to be tested across demographic groups to ensure it doesn't perpetuate inequality.
The goal of AI in citizen services is to make the human touch count more, not less — by directing it where it genuinely matters.
A Practical Starting Point
For government entities considering AI, the practical starting point is internal — not citizen-facing. Start with AI that helps public servants do their jobs: search across internal knowledge bases, summarize regulatory documents, draft routine correspondence, and flag inconsistencies in application data. The benefits are immediate, the risks are lower, and the internal confidence built from these tools creates the foundation for citizen-facing deployment.