Revolution in Archaeology: Virtual Excavation with AI iDig
Artificial intelligence
The days of an archaeologist studying soil traces with a shovel in hand seem to be numbered. Digital archaeologists will soon launch the first version of software capable of carrying out a virtual excavation.
Virtual excavation assistant
Excavations are time-consuming, expensive, and disruptive. How much better would it be to conduct them entirely virtually? With that idea in mind, digital archaeologists from Leiden University set out to develop an archaeological virtual assistant. 'This is the ultimate non-invasive excavation technique!'
'The biggest challenge was bringing together a wide range of different sources', explains team leader and professor C.O. Pilot. 'A virtual archaeologist must first understand the composition of the soil. Is it sand? Clay? Peat? That information determines what you might expect to find.' To this end, the AI, named iDig, was given full access to the geological map of the Netherlands.
Predictions
The next step was predicting what kind of archaeological materials might surface during an excavation. 'For a virtual excavation on the Veluwe, for example, that could be Bronze Age and Iron Age materials, though we also know there’s a small chance of a Roman marching camp', adds image specialist M.I.D. Journey.
'Based on what has previously been found locally and the soil type in which the virtual excavation takes place, iDig predicts what will be found at that specific location.' Journey presses a button, and iDig instantly produces a stratigraphic overview of a new virtual excavation. 'Look, it’s just found a completely unexpected third Roman marching camp near Ermelo. Incredible.'
Concerns
Wherever AI is developed, concerns tend to follow. 'We’re learning all kinds of field techniques right now', says archaeology student Claude, 'but there’s a good chance we’ll never actually use them. I’ve heard that iDig is already faster and more effective than a team of experienced archaeologists, and it’s still in the testing phase!'
Pilot understands the concerns. 'The archaeologist of the future will be more of a programmer than a field technician. There’s no way around that. But the current generation of students will certainly still get their boots in the mud. The system is nowhere near good enough to replace a fully trained archaeologist.'
And how does the professor respond to questions about sustainability and artificial intelligence? 'It’s true that AI consumes a lot of energy, but remember that archaeological excavations do too, just think of the travel movements of an excavation team. My suspicion is that archaeology will become more sustainable if we switch fully to virtual excavation.'
Data scraping
One of iDig’s growing pains is that it still struggles with fictional information. For example, it generated an archaeological model of a Gallic village in Brittany surrounded by Roman military camps. 'A classic case of failed data scraping', Pilot shrugs. 'These are exactly the kinds of issues that come up during the testing phase.' She laughs. 'Did you know that iDig performs a virtual excavation more slowly when it takes place in clay soil? It taught itself that digging in clay takes more effort.'
Prehistoric village
In the computer lab, a group of students hunch over a virtual excavation. 'iDig just found a prehistoric village under Leiden’s city hall! So cool, you’d never be able to excavate there yourself. And now we actuall know what’s down there.' iDig presents the various Iron Age farmhouses in all their polished beauty. 'Now all I have to do is feed the virtual excavation data into Chat, and I’ll have a thesis.' The fact that the iDig model shows the prehistoric inhabitants of Leiden eating hutspot doesn’t seem to bother them.