Large Language Models and Generative AI, Oh My! | Advances in Archaeological Practice | Cambridge Core
| Source: Mastodon | Original article
Peter Cobb’s new essay, “Large Language Models and Generative AI, Oh My!”, appears in Cambridge Core’s Advances in Archaeological Practice Volume 11, Special Issue 3, and maps the rapid infiltration of tools such as ChatGPT, Midjourney and emerging multimodal models into archaeological research. Cobb argues that generative AI is already reshaping fieldwork documentation, artifact classification and the drafting of excavation reports, while also surfacing a suite of ethical dilemmas that the discipline has yet to resolve.
The piece catalogues concrete experiments: LLM‑driven transcription of epigraphic corpora, image‑to‑text pipelines that suggest typologies for pottery shards, and automated narrative generation that can turn raw field notes into publishable prose within minutes. Proponents cite speed gains, lower barriers for scholars in under‑funded institutions, and the potential to synthesize disparate datasets across regions. Critics, however, warn that black‑box models may propagate biases embedded in training data, obscure provenance, and encourage a “plug‑and‑play” mindset that sidelines critical interpretation. Cobb stresses that archaeological heritage—often tied to indigenous and contested histories—requires transparent provenance tracking and consent mechanisms that current AI platforms rarely provide.
Why it matters now is twofold. First, the sheer scale of LLMs means that even niche domains like archaeology can tap into massive linguistic and visual knowledge bases without building bespoke models. Second, the discipline’s methodological rigor makes it a litmus test for how humanities fields can adopt AI responsibly, balancing acceleration with stewardship of cultural memory.
Looking ahead, the community should watch for the rollout of domain‑specific LLMs trained on curated archaeological corpora, the formation of ethical guidelines by bodies such as the European Association of Archaeologists, and upcoming workshops at the International Congress of Archaeological Sciences that will benchmark AI‑augmented workflows. The next wave of funding calls from the EU’s Horizon Europe programme is also likely to prioritize projects that couple generative AI with heritage preservation, setting the agenda for how the field navigates this technological crossroads.
Sources
Back to AIPULSEN