Adnan KhanAdnanProduct & AI
San Francisco Bay Area
Get in touch

Source note3 min read

The Context Access Divide: Interaction-Level Architecture as a Complementary Dimension of Agentic Inequality

Source

arXiv

arxiv.org/abs/2607.08495v1

Read the source

Source

Plain-Language Brief

The text introduces a concept called 'Context Access Divide' in the field of AI, which refers to a qualitative difference in AI utility experienced by two users with similar access levels. This divide arises when one system can autonomously retrieve context from the user's knowledge corpus while another requires manual intervention. The author proposes 'contextuality' as a new dimension of AI-mediated inequality and offers a probabilistic model to demonstrate this divide, drawing on literature in cognitive psychology.

What the Document Says

  • The text discusses the Context Access Divide (CAD) as a divide in AI utility for users with similar access levels based on whether the system can autonomously retrieve context or requires manual intervention.
  • It proposes contextuality as a new dimension of AI-mediated inequality, complementing but not reducible to the Sharp et al. framework.
  • A probabilistic model grounded in fan effect literature is introduced to demonstrate the CAD and its implications for knowledge-work stratification and AI platform governance.

Key Concepts Mentioned

  • Context Access Divide
  • contextuality
  • AI-mediated inequality
  • knowledge-intensive workers
  • Dynamic Context Retrieval
  • Manual Attachment

Claims Made by the Document

  • The CAD constitutes a qualitative threshold in AI usefulness for knowledge-intensive workers.
  • Manual context attachment leads to a combinatorial collapse in task-success probability as corpus size and task conjunctivity grow.

Evidence Used

  • The text provides a probabilistic model grounded in fan effect literature to demonstrate the CAD.

Methods / Approach

  • Probabilistic modeling based on fan effect literature

Limitations Stated by the Authors

Not stated in source

Limitations Noted by the Agent

Agent interpretation — not from the source.

The provided text does not specify any limitations of the author's work.

Open Questions

  • What is the fan effect literature referenced in the text?

Notes for Future Q&A

May include interpretation. Labeled where it does.

  • Sharp et al. (2025) is likely a reference to a previous study on agentic inequality mentioned but not cited directly in this text.
  • The contextuality dimension of AI-mediated inequality may have implications for the design and development of future AI systems.

Raw Abstract / Excerpt

Sharp et al. (2025) introduce 'agentic inequality' as a framework for analyzing disparities in access to AI agents across three dimensions: availability, quality, and quantity. These person- and organization-level dimensions characterize who can access agents and at what capability, but do not address a structurally important divide operating at a finer level: the individual interaction. Two users with nominally equivalent agent access may experience qualitatively different AI utility depending on whether the system can autonomously retrieve context from the user's knowledge corpus (Dynamic Context Retrieval) or requires the user to manually identify and attach relevant documents at each query (Manual Attachment). We term this the Context Access Divide (CAD). For knowledge-intensive workers whose intellectual capital spans tens of thousands of files, the CAD constitutes a qualitative threshold in AI usefulness: below it, the cognitive burden of context curation falls on the human, reproducing the inefficiencies AI is meant to eliminate. We propose contextuality -- the degree to which an AI system autonomously accesses a user's accumulated knowledge capital -- as a dimension of AI-mediated inequality that complements, but is not reducible to, the Sharp et al. framework. We formalize the CAD with a probabilistic model grounded in the fan effect literature in cognitive psychology, demonstrating that manual context attachment leads to a combinatorial collapse in task-success probability as corpus size and task conjunctivity grow, while dynamic retrieval architectures are structurally insulated from this collapse. We analyze the technical basis of this divide in the Model Context Protocol (MCP) and retrieval-augmented generation (RAG) architectures, and examine its implications for knowledge-work stratification and AI platform governance.