---
id: ai-for-archaeology
title: "AI for Archaeology: Site Detection, Artifact Classification, and Digital Heritage Preservation"
schema_type: article
category: ai
language: en
confidence: medium
last_verified: "2026-05-28"
created_date: "2026-05-24"
generation_method: ai_structured
ai_models:
  - claude-4.5-sonnet
derived_from_human_seed: true
conflict_of_interest: none_declared
is_live_document: false
data_period: static
completeness: 0.85
atomic_facts:
  - id: af-ai-for-archaeology-1
    statement: >-
      A 2024 PNAS paper reported an AI-assisted survey that discovered 303 new figurative geoglyphs
      in the Nazca Pampa during a six-month field survey.
    source_title: >-
      AI-accelerated Nazca survey nearly doubles the number of known figurative geoglyphs and sheds
      light on their purpose
    source_url: https://www.pnas.org/doi/10.1073/pnas.2407652121
    confidence: medium
  - id: af-ai-for-archaeology-2
    statement: >-
      The Nature paper on Ithaca described a deep neural network for restoring missing text and
      assisting geographic and chronological attribution of ancient Greek inscriptions.
    source_title: Restoring and attributing ancient texts using deep neural networks
    source_url: https://www.nature.com/articles/s41586-022-04448-z
    confidence: medium
  - id: af-ai-for-archaeology-3
    statement: >-
      The National Endowment for the Humanities reported that Vesuvius Challenge researchers used
      machine-learning algorithms to detect and decipher passages from scans of carbonized
      Herculaneum scrolls.
    source_title: Students Decipher 2,000-Year-Old Herculaneum Scrolls
    source_url: https://www.neh.gov/news/students-decipher-2000-year-old-herculaneum-scrolls
    confidence: medium
primary_sources:
  - id: ps-ai-for-archaeology-1
    title: >-
      AI-accelerated Nazca survey nearly doubles the number of known figurative geoglyphs and sheds
      light on their purpose
    type: journal_article
    year: 2024
    institution: Proceedings of the National Academy of Sciences
    doi: 10.1073/pnas.2407652121
    url: https://www.pnas.org/doi/10.1073/pnas.2407652121
  - id: ps-ai-for-archaeology-2
    title: Restoring and attributing ancient texts using deep neural networks
    type: journal_article
    year: 2022
    institution: Nature
    doi: 10.1038/s41586-022-04448-z
    url: https://www.nature.com/articles/s41586-022-04448-z
  - id: ps-ai-for-archaeology-3
    title: Students Decipher 2,000-Year-Old Herculaneum Scrolls
    type: institutional_article
    year: 2024
    institution: National Endowment for the Humanities
    url: https://www.neh.gov/news/students-decipher-2000-year-old-herculaneum-scrolls
known_gaps:
  - >-
    Multi-modal fusion of geophysical survey data with satellite imagery for comprehensive
    subsurface site mapping
  - Standardized open-access archaeological datasets for reproducible AI benchmarking
disputed_statements: []
secondary_sources: []
updated: "2026-05-28"
---
## TL;DR
AI helps archaeology when it accelerates specific expert workflows: scanning imagery for site candidates, restoring damaged inscriptions, and detecting text in otherwise unreadable heritage materials.

## Core Explanation
The best-supported examples are narrow and human-supervised. AI can prioritize candidate geoglyphs for field validation, assist historians with damaged inscriptions, and support computer-vision challenges around carbonized scrolls. These tools aid archaeological interpretation; they do not replace excavation, dating, provenance work, or specialist review.

## Related Articles

- [AI for Cultural Heritage: Digital Preservation, Art Attribution, and Museum Intelligence](../ai-for-cultural-heritage.md)
- [AI for Astronomy: Exoplanet Detection, Galaxy Classification, and Computational Astrophysics](../ai-for-astronomy.md)
- [3D Human Modeling: Parametric Body Models, Mesh Recovery, and Digital Avatars](../3d-human-modeling.md)