AI-Accelerated Site Discovery from Imagery: Yamagata + IBM's Nazca Pipeline

AI-Accelerated Site Discovery from Imagery: Yamagata + IBM's Nazca Pipeline

The most quantitatively successful AI site-discovery pipeline in archaeology keeps compounding, and it runs on a corporate research lab as technical patron. The primary publication — Sakai et al., "AI-accelerated Nazca survey nearly doubles the number of known figurative geoglyphs and sheds light on their purpose" (PNAS, Sept 24, 2024) — states: "we report the deployment of an AI system to the entire Nazca region, a UNESCO World Heritage site, leading to the discovery of 303 new figurative geoglyphs within only 6 mo of field survey, nearly doubling the number of known figurative geoglyphs." The AI flagged 1,309 candidate locations; field teams walked roughly a quarter of them (1,440 labor hours) and confirmed 303 — which the authors call "another 16-fold acceleration... using big geospatial data technologies and data mining with the aid of AI."

The player structure is the key signal: this is a joint project of Yamagata University's Institute of Nasca and IBM Research. Per Yamagata's press release: "IBM Research developed an AI model which is able to work with just a few training samples while still being very performant." Few-shot learning is exactly the constraint archaeology imposes — there are never many labeled examples of a site class.

Fresh feature coverage (Futura-Sciences, resurfaced Oct 4, 2026) puts the running total at 893 confirmed geoglyphs, 781 of them AI-assisted (303 in the PNAS paper + 248 more announced at Expo 2025 Osaka), and quotes project leader Masato Sakai: "The traditional method — visually identifying geoglyphs from high-resolution images of this vast area — was slow and carried the risk of overlooking them." Futura adds that ~968 AI-flagged sites remain unassessed and the team's Peru permit runs through 2026¹, with the final unmapped stretches of the Pampa next.

What it means

• Players: Yamagata Institute of Nasca + IBM Research is now the flagship pairing for imagery-based discovery. IBM's contribution mirrors the patron pattern in decipherment (ÖAW + Mistral's Apollo, see Commercial CRM and Funding: Corporate AI Patronage Arrives — a National Academy's Greek Model for €400K): a corporate AI lab supplies model-building capacity, not just cash, and gets a PNAS-grade proof point.
• The workflow is "algorithm proposes, human disposes": 1,309 candidates → 303 confirmed by walking the ground. As Futura puts it: "The algorithm provides the lead; the shovel provides the proof."
• Business angle: few-shot object detection re-pointed at existing high-resolution imagery is cheap to transfer to other deserts/trailscapes — a service opportunity distinct from the hardware side (LiDAR-as-a-service, see Subsurface and Remote Archaeology: Yamaha Brings Forestry LiDAR-as-a-Service to Copán). The unverified backlog (~968 sites) is itself a market: someone gets paid to walk it.