Artificial Intelligence (AI) is reshaping the landscape of educational assessment, offering opportunities to improve accuracy, efficiency, and learner personalization. This narrative review explores recent empirical literature to evaluate the implementation, challenges, and global perspectives of AI in educational assessments. The review synthesizes findings from a comprehensive literature search across major academic databases, including Scopus, PubMed, and Google Scholar, using targeted keywords related to AI-enhanced assessment. Studies were selected based on inclusion criteria emphasizing relevance to AI-driven assessment systems across diverse educational contexts. Results indicate that AI significantly improves assessment accuracy, reduces educator workload, and enhances learner engagement through adaptive feedback. Comparative evidence shows high congruence between AI-generated and human grading outcomes, with students responding positively to real-time feedback systems. Global case studies from the United States, Finland, China, and India illustrate varying approaches to AI integration, shaped by policy, infrastructure, and cultural context. However, ethical concerns such as algorithmic bias, data privacy, and the diminishing role of human judgment remain persistent barriers. The discussion highlights the importance of ethical frameworks, professional development, and inclusive policies to ensure equitable AI implementation. This review underscores the need for proactive strategies, including bias audits and stakeholder collaboration, to address existing challenges. By aligning AI tools with pedagogical goals and ethical standards, educational institutions can leverage AI to create more effective and inclusive assessment practices.