Abstract
Memory is a fundamental component of AI systems, underpinning large languagemodels (LLMs) based agents. While prior surveys have focused on memoryapplications with LLMs, they often overlook the atomic operations that underliememory dynamics. In this survey, we first categorize memory representationsinto parametric, contextual structured, and contextual unstructured and thenintroduce six fundamental memory operations: Consolidation, Updating, Indexing,Forgetting, Retrieval, and Compression. We systematically map these operationsto the most relevant research topics across long-term, long-context, parametricmodification, and multi-source memory. By reframing memory systems through thelens of atomic operations and representation types, this survey provides astructured and dynamic perspective on research, benchmark datasets, and toolsrelated to memory in AI, clarifying the functional interplay in LLMs basedagents while outlining promising directions for future research\footnote{Thepaper list, datasets, methods and tools are available at\href{https://github.com/Elvin-Yiming-Du/Survey_Memory_in_AI}{https://github.com/Elvin-Yiming-Du/Survey\_Memory\_in\_AI}.}.