EXPLAINABLE AIOPS: A DEEP SURVEY ON TRUSTWORTHY AND TRANSPARENT AI IN CLOUD-SCALE DEVOPS AUTOMATION

Abstract

As Artificial Intelligence for IT Operations (AIOps) becomes increasingly integral to managing modern, cloud-scale DevOps environments, concerns about the opacity of AI-driven decisions have grown significantly. The adoption of black-box models in these systems, while enabling rapid automation, introduces critical challenges in trust, auditability, and regulatory compliance. This undermines confidence in automated decisions, especially in high-stakes environments involving anomaly detection, root cause analysis, and predictive scaling. To address these challenges, the integration of Explainable Artificial Intelligence (XAI) into AIOps has emerged as a key research and industrial imperative. XAI techniques promise to make AI behaviors transparent, interpretable, and human-understandable, thus facilitating trust and control in AI-augmented operational workflows. However, the landscape of explainable AIOps is fragmented across tools, methods, and deployment scenarios, and there is no consolidated resource that comprehensively maps the field. This survey presents a deep, structured analysis of over 70 scholarly works spanning academic literature and industry applications. We develop a taxonomy of explainable AIOps techniques, categorizing them by method type (e.g., model-agnostic, deep learning-based, symbolic), DevOps use-case (e.g., monitoring, RCA, auto-remediation), and cloud-specific integration. Through this analysis, we highlight significant gaps in scalability, standardization, and usability that current approaches fail to address in dynamic cloud environments. Finally, the paper provides a roadmap for future research directions, including hybrid neuro-symbolic explainability, human-in-the-loop systems, and edge-cloud trustworthy AIOps. This work aims to serve as a foundational reference for researchers and practitioners seeking to build transparent, trustworthy, and scalable AI systems for modern cloud operations.