An artificial intelligence challenge to reassemble nearly 20,000 fragments of 3,000-year-old pottery recently demonstrated a gap between AI's archaeological promise and its current capabilities.

The challenge, launched in 2024 by Shandong University and Beijing Normal–Hong Kong Baptist University, asked participants to develop algorithms capable of piecing together pottery fragments degraded by three millennia of weathering, abrasion, and color distortion.

The results, released on Saturday at a forum in Qingdao, Shandong province, showed that only 78 fragment pairs were matched, highlighting persistent technical bottlenecks in applying AI to complex archaeological tasks.

The thin results have not dampened ambitions. Fang Hui, dean of the Institute of Cultural Heritage at Shandong University, said a university research team has since matched more than 250 fragment pairs using a newer reconstruction framework and predicted AI would eventually rank alongside radiocarbon dating and ancient DNA analysis as a transformative force in the field.

"The forms and quantities of reassembled artifacts can offer critical clues to understanding social organization, daily life, and cultural exchanges in ancient times," Fang said.

Others held a more measured view of where the technology currently stands. Zhichun Jing, an associate professor of anthropology at the University of British Columbia in Canada, said deep learning has improved efficiency in numerous ways, but remains limited to materials already known to researchers.

He argued the more significant opportunity lies in what he called "active inference", which shifts the focus from faster recognition and classification to guiding what evidence should be collected next and how.

In archaeological contexts marked by incomplete evidence and competing interpretations, he said the approach emphasizes identifying the most informative observations and materials that could reshape existing understanding.

The forum drew more than 260 experts in archaeology, computer science, and cultural heritage from China and abroad.