AI Misidentifies Maya Site Already Documented by Experts

A recent incident involving Google's Pietro Miotti has sparked a significant debate within the archaeological community. Miotti claimed to have discovered a new Maya site in Mexico using AI agents, only to be informed that the site had already been documented in an academic journal over a year ago.

Origins of the AI Exploration

During a trip to Peru, Miotti pondered the possibilities modern technology could provide to archaeologists and historical explorers. This line of inquiry led him to utilize lidar technology, which employs laser mapping to generate intricate three-dimensional renderings of landscapes. After assembling a team of 48 AI agents, he leveraged NASA’s lidar data to search for undocumented archaeological sites. However, he soon shifted his focus to the Maya region, citing limited public data availability in Peru.

AI Findings and Initial Claims

According to Miotti, the AI agents identified several potential locations, one of which appeared to be an unregistered village in the existing archaeological records. In his rush to share his findings, he contacted Mexico's National Institute of Anthropology and History (INAH) using AI translation tools to facilitate communication in Spanish. He reported that INAH initially did not have a registered site at this location.

Miotti expressed excitement at the rapid response he received, declaring it a remarkable demonstration of how technology can transform science. However, this claim quickly drew the ire of established archaeologists.

Criticism from Archaeological Experts

Social media platforms were rife with responses from archaeologists criticizing Miotti's oversight. Sarah Parcak, an archaeology professor, highlighted the pressing need for thorough examination against existing research. She pointed out that a lack of registration in government databases does not equate to a site being unknown. Her sentiments were echoed by other professionals in the field who condemned the potential misrepresentation and crediting of unqualified efforts over established researchers.

Miotti, acknowledging his lack of formal training in archaeology, expressed regret over the misunderstanding. He emphasized that comprehensive expertise is crucial for verifying such AI-derived findings. According to him, the inherent limitations of AI meant that discerning the nuances of archaeological data—especially with regard to previously published work—is complex and often beyond these agents' capabilities.

The Role of AI in Future Research

Despite facing backlash, Miotti remains optimistic about the future potential of AI in supporting archaeological research. He cited encouraging feedback from various archaeologists and doctoral students eager to integrate these technologies into their methodologies.

However, Elizabeth Graham, a leading Mesoamerican archaeology expert, underscored that the real challenges in archaeological work often stem from the scarcity of funding for actual site exploration rather than the discovery of new sites. Many sites are likely already known to specialists but are underfunded and unexcavated.

Broader Implications and Concerns

Ivan Šprajc, another archaeology professor, noted that the archaeological community has been utilizing lidar and other remote sensing technologies effectively long before this incident. He expressed concern that the rising popularity of amateur AI-assisted archaeology could inundate established researchers with claims of new discoveries that ultimately do not advance scientific knowledge.

Moreover, Šprajc warned about the risks of publishing locations of archaeological sites, which could inadvertently draw looters. Despite these dangers, he doesn't view the current situation as critical unless it leads to theft of artifacts.

This episode highlights ongoing challenges at the intersection of AI and traditional scientific fields. While AI presents exceptional capabilities, its application by non-experts raises issues of accuracy and interpretation of established findings. The experience serves to remind the scientific community of the vital necessity for collaboration between experts and emerging technologies, ensuring that the full potential of advancements like AI is harnessed effectively and responsibly.