GENERATIVE AI-ENHANCED IT SERVICE MANAGEMENT FOR CLOUD OPERATIONS IN SAUDI GOVERNMENT ORGANIZATIONS

Generative artificial intelligence is moving IT service management from retrospective ticket administration towards conversational diagnosis, knowledge synthesis and partially automated remediation. Its use in Saudi government cloud operations is nevertheless a high-consequence socio-technical intervention: operational telemetry is sensitive, service interruption can affect citizens, Arabic and English records are heterogeneous, and plausible but incorrect advice can amplify failure. This narrative review critically synthesises thirty peer-reviewed journal articles published from 2020 to 2025. It examines incident intelligence, retrieval-grounded generation, observability correlation, agentic automation, human oversight and public-sector governance, while interpreting their relevance to Saudi digital-government conditions. Evidence was selected through database searching, quality screening and thematic comparison rather than statistical aggregation. The synthesis finds that discriminative models remain preferable for high-frequency detection and classification, whereas generative models add most value in semantic triage, cross-source explanation, runbook adaptation and knowledge capture. The strongest architecture is therefore layered: sovereign data controls and an event fabric support specialised analytics; retrieval constrains a generative reasoning service; policy gates, simulation and human approval control action. Published accuracy results are not directly transferable because datasets, labels and operational definitions differ, and few studies evaluate bilingual, cross-agency or long-running deployments. A risk-tiered adoption pathway is proposed, linking autonomy to evidence quality, reversibility and service criticality. The review identifies research priorities in Arabic operational language, benchmark design, causal diagnosis, continuous assurance, cost-aware inference and measurable public value. It concludes that trustworthy augmentation, rather than model novelty or maximum autonomy, should define generative AI-enabled service management under Saudi digital transformation.