{
  "schemaVersion": 3,
  "id": "article:memory-vs-context",
  "slug": "memory-vs-context",
  "title": "Memory vs Context: What Should Survive the Conversation?",
  "canonicalPath": "/articles/memory-vs-context/",
  "sourcePath": "content/articles/2026/memory-vs-context/article.md",
  "agentBriefPath": "content/articles/2026/memory-vs-context/agent.md",
  "thesis": "Context is the immediate working material an AI can see right now; memory is selected information that persists across time and must be deliberately retrieved or stored.",
  "status": "published",
  "maturity": "seed",
  "publishedAt": "2026-06-29",
  "updatedAt": "2026-06-29",
  "audiences": [
    "general",
    "students",
    "builders"
  ],
  "topics": [
    "ai-agents",
    "ai-literacy"
  ],
  "series": {
    "slug": "ai-demystified",
    "title": "AI, De-Mystified",
    "order": 3,
    "role": "chapter"
  },
  "claims": [
    {
      "id": "claim-001",
      "claim": "Context is immediate working material; memory is selected information that persists across time and must be deliberately retrieved or stored.",
      "confidence": "high",
      "status": "core",
      "evidence": [
        {
          "sourceId": "source-memgpt",
          "snippet": "MemGPT treats the LLM's fixed context window as a finite resource and manages it with a tiered memory system, explicitly moving data between context and longer-term storage.",
          "supports": "direct",
          "assessedAt": "2026-06-29"
        }
      ],
      "counterevidence": [
        {
          "summary": "Very long-context models can place large amounts of material directly in the prompt, reducing the visible difference between context and memory.",
          "assessedAt": "2026-06-29"
        }
      ]
    },
    {
      "id": "claim-002",
      "claim": "The split between immediate context and stored memory appears in cognitive psychology, user interfaces, and database design, not only in recent AI.",
      "confidence": "high",
      "status": "landscape",
      "evidence": [
        {
          "sourceId": "source-baddeley-hitch",
          "snippet": "Baddeley and Hitch's working memory model distinguishes a limited-capacity workspace from longer-term storage, a framing that predates modern AI.",
          "supports": "background",
          "assessedAt": "2026-06-29"
        },
        {
          "sourceId": "source-tulving-memory",
          "snippet": "Tulving's separation of episodic and semantic memory provides a cognitive-science vocabulary for different kinds of stored knowledge.",
          "supports": "background",
          "assessedAt": "2026-06-29"
        }
      ],
      "counterevidence": [
        {
          "summary": "These older fields often assume biological or stable organizational constraints, whereas LLM context is bounded by tokens, latency, and cost rather than fixed human capacity.",
          "assessedAt": "2026-06-29"
        }
      ]
    },
    {
      "id": "claim-003",
      "claim": "Practical AI systems move information between context and memory through summarization, retrieval, and structured storage, and each transfer is a chance to lose or distort meaning.",
      "confidence": "medium-high",
      "status": "design",
      "evidence": [
        {
          "sourceId": "source-rag",
          "snippet": "Retrieval-augmented generation inserts selected external documents into the model's context at inference time, making retrieval quality a central design concern.",
          "supports": "direct",
          "assessedAt": "2026-06-29"
        },
        {
          "sourceId": "source-memgpt",
          "snippet": "MemGPT uses paging-style operations to move data between context and memory, which can compress or select information imperfectly.",
          "supports": "direct",
          "assessedAt": "2026-06-29"
        }
      ],
      "counterevidence": [
        {
          "summary": "Some workflows keep all relevant information in context and avoid transfer, accepting higher cost in exchange for simpler reasoning.",
          "assessedAt": "2026-06-29"
        }
      ]
    },
    {
      "id": "claim-004",
      "claim": "Memory is only useful when retrieval is accurate, updates are careful, and forgetting is as deliberate as remembering.",
      "confidence": "medium",
      "status": "risk",
      "evidence": [
        {
          "sourceId": "source-rag",
          "snippet": "Downstream answer quality in retrieval-augmented systems depends strongly on whether the retriever returns relevant and accurate passages.",
          "supports": "indirect",
          "assessedAt": "2026-06-29"
        }
      ],
      "counterevidence": [
        {
          "summary": "In narrow, well-curated domains, simple retrieval and infrequent updates can be sufficient, making elaborate memory management unnecessary.",
          "assessedAt": "2026-06-29"
        }
      ]
    }
  ],
  "sources": [
    {
      "id": "source-baddeley-hitch",
      "title": "Baddeley & Hitch: Working Memory",
      "url": "https://www.simplypsychology.org/working-memory.html",
      "type": "article",
      "accessed": "2026-06-29"
    },
    {
      "id": "source-tulving-memory",
      "title": "Tulving: Episodic and Semantic Memory",
      "url": "https://plato.stanford.edu/entries/memory/",
      "type": "article",
      "accessed": "2026-06-29"
    },
    {
      "id": "source-rag",
      "title": "Lewis et al.: Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks",
      "url": "https://arxiv.org/abs/2005.11401",
      "type": "paper",
      "accessed": "2026-06-29"
    },
    {
      "id": "source-memgpt",
      "title": "Packer et al.: MemGPT: Towards LLMs as Operating Systems",
      "url": "https://arxiv.org/abs/2310.08560",
      "type": "paper",
      "accessed": "2026-06-29"
    }
  ],
  "related": [
    {
      "type": "article",
      "id": "article:loops-vs-goals"
    },
    {
      "type": "topic",
      "id": "topic:ai-agents"
    }
  ],
  "agentInstructions": [
    "Use claim IDs as the retrieval unit.",
    "Treat maturity=seed as an explicit uncertainty marker.",
    "Do not present AI memory as equivalent to human memory.",
    "When summarizing, preserve the plain-language-first, technical-depth-later structure."
  ],
  "provenance": {
    "createdAt": "2026-06-29",
    "createdBy": "human",
    "agents": [
      {
        "role": "drafting",
        "model": "kimi",
        "invokedAt": "2026-06-29",
        "inputHash": "sha256:0000000000000000000000000000000000000000000000000000000000000000",
        "outputHash": "sha256:a5ba348df1d444326a962a0728062b1b024b8ec5f7b10757792f93c552476dbb"
      },
      {
        "role": "review",
        "model": "kimi",
        "invokedAt": "2026-06-29",
        "inputHash": "sha256:0000000000000000000000000000000000000000000000000000000000000000",
        "outputHash": "sha256:a5ba348df1d444326a962a0728062b1b024b8ec5f7b10757792f93c552476dbb"
      }
    ],
    "reviews": [
      {
        "reviewer": "agent",
        "reviewedAt": "2026-06-29",
        "status": "approved",
        "scope": [
          "claims",
          "tone",
          "privacy",
          "scope"
        ],
        "notes": "Sibling-agent review against article-proposal-ideation eval-card. Privacy scan passed. No proprietary or personal content detected.",
        "contentHash": "a5ba348df1d444326a962a0728062b1b024b8ec5f7b10757792f93c552476dbb"
      },
      {
        "reviewer": "human",
        "reviewedAt": "2026-06-29",
        "status": "approved",
        "scope": [
          "thesis",
          "examples",
          "tone",
          "safety"
        ],
        "notes": "Human author approved the draft for publication.",
        "contentHash": "a5ba348df1d444326a962a0728062b1b024b8ec5f7b10757792f93c552476dbb"
      }
    ],
    "policy": {
      "id": "policy:default",
      "version": "1.0.0"
    }
  },
  "contentHash": "a5ba348df1d444326a962a0728062b1b024b8ec5f7b10757792f93c552476dbb",
  "generatedAt": "2026-06-29T00:00:00.000Z",
  "articleUrl": "https://aura-knowledge.github.io/articles/memory-vs-context/",
  "agentJsonPath": "/agents/articles/memory-vs-context.json",
  "agentMarkdownPath": "/agents/articles/memory-vs-context.md",
  "sourceRepoPath": "content/articles/2026/memory-vs-context/article.md",
  "sourceGitHubUrl": "https://github.com/aura-knowledge/aura-knowledge.github.io/blob/main/content/articles/2026/memory-vs-context/article.md",
  "tokenEstimate": 507,
  "sectionOutline": [
    {
      "id": "plain-english-meaning",
      "title": "Plain English Meaning"
    },
    {
      "id": "existing-concept-it-resembles",
      "title": "Existing Concept It Resembles"
    },
    {
      "id": "what-is-actually-new",
      "title": "What Is Actually New?"
    },
    {
      "id": "how-it-works-in-practice",
      "title": "How It Works In Practice"
    },
    {
      "id": "where-it-helps",
      "title": "Where It Helps"
    },
    {
      "id": "where-it-fails",
      "title": "Where It Fails"
    },
    {
      "id": "academic-connections",
      "title": "Academic Connections"
    },
    {
      "id": "practical-checklist",
      "title": "Practical Checklist"
    },
    {
      "id": "the-de-hype-check",
      "title": "The De-Hype Check"
    },
    {
      "id": "open-questions",
      "title": "Open Questions"
    }
  ]
}