The neural network connected two of the human analysts with accuracy and beat the other two, the researchers found.

The machine was also far more efficient. Because the task was boring, none of the human analysts wanted to go through all 3,000 photos without stopping, said Dr. Pawlowicz. Although they could probably have completed the task in three hours, each did the analysis in multiple sessions over three to four months.

The neural network whipped through thousands of images in a matter of minutes.

Not only was the computer program more efficient and accurate than the archaeologists, it was better at articulating why it had categorized broken glass in a certain way compared to its living, breathing competitors. In one case, the computer provided a new intelligent sorting observation for the researchers: it indicated that two similar types of ceramics with design elements with barbed lines can be distinguished by whether the lines are connected at right angles or parallel, said Leszek Pawlowicz, an additional faculty member at Northern Arizona University and another author on the study.

The machine also outperformed humans by only offering one answer for each classification. Participating archaeologists often disagreed on how items were categorized, a known problem that often slows archaeological projects, the authors said.

Phillip Isola, a professor of electrical engineering and computer science at MIT, who was not involved in the study, said he wasn’t surprised that the neural network did as well, or sometimes better, than the archaeologists.

“It’s the same story we’ve heard a couple of times now,” said Dr. Isola. In the medical imaging field, for example, researchers have found that neural networks can compete with radiologists in identifying tumors. Academics also use similar tools to categorize plant and bird types.

This is also far from the first time archaeologists have turned to artificial intelligence. In 2015, researchers in France used machine learning to classify medieval French ceramics. A group of archaeologists and computer scientists from five countries is currently developing a digital tool for categorizing ceramic shards. However, none of these projects explicitly put people against machines.