LLMography: Transforming Human–AI Conversations into Traceability, Oversight, and Auditability Indicators

Abstract

The growing use of large language models in education, software engineering, academic writing, and technical documentation raises a question that output detection cannot answer: how was a result produced? This paper introduces LLMography as a framework for transforming Human–AI conversations into indicators of provenance, human contribution, AI dependency, reproducibility, and auditability. By analogy with bibliography and webography, it treats conversation history as a structured trace of co-production rather than a list of prompts.

A prototype analyzes conversation traces and produces reports including prompt quality, human direction, AI dependency, auditability, output traceability, privacy risk, and a recommended LLMography label. A preliminary exploratory evaluation on 19 anonymized student audit reports is reported in the preprint. The manuscript also applies the framework to its own writing process.

Key contributions

1. Introduces LLMography as a term and computational framework.
2. Positions the idea by analogy with bibliography and webography, then extends it to dynamic co-construction.
3. Defines conversation-level indicators for direction, dependency, traceability, and auditability.
4. Presents a prototype reporting pipeline.
5. Reports a small exploratory evaluation.
6. Applies the method to the paper’s own production.

Research figures

Figures from the preprint should be linked from the PDF. They are not reproduced here as decorative assets.

Note on later work

The conversation-level prototype is a foundation. Later LLMography work extends the same reconstruction problem to loops, agents, tools, and graphs.

Bousmah, M. (2026). LLMography: Transforming Human–AI Conversations into Traceability, Oversight, and Auditability Indicators. arXiv:2606.29437.

@misc{llmographyhumanaiconversations,
  title = {LLMography: Transforming Human–AI Conversations into Traceability, Oversight, and Auditability Indicators},
  author = {Mohammed Bousmah},
  year = {2026},
  eprint = {2606.29437},
  doi = {10.48550/arXiv.2606.29437},
  url = {https://llmography.ai/research/papers/llmography-human-ai-conversations}
}