Over the past decade, the use of Artificial Intelligence (AI) has become increasingly common across various fields of science and technology. Archaeology, too, has begun to explore the potential of AI, with dedicated sessions now focusing on its diverse applications within the discipline.

Archaeology benefits significantly from AI, particularly when there is a need to analyse large volumes of data or perform complex, highly specialised, and time-consuming tasks. Currently, AI is applied to many archaeological fields such as:

• Artifact Identification: AI assists in classifying and dating pottery shards through image recognition;
• Site Detection: Machine Learning analyses satellite images to discover hidden archaeological sites;
• Language Decoding: AI assists in deciphering ancient scripts like Linear B or cuneiform;
• 3D Reconstruction: AI reconstructs damaged statues or buildings from fragmented remains; and
• Predictive Modelling: AI predicts where undiscovered artifacts may be based on past data.

AI offers tremendous potential to advance our understanding of shared archaeological heritage. However, several critical issues now demand closer attention. These include identifying specific archaeological research questions that AI is best suited to address. Ensuring access to appropriate data, both its availability and the means to generate it effectively. Navigating ethical, epistemological, and interpretive challenges posed by AI in archaeological contexts. Addressing the lack of sustainable infrastructure and resources necessary to support this work. Each of these areas is essential and requires deeper, focused discussion and exploration.

Despite popular perception, one of the greatest challenges in AI is not developing suitable algorithms, but rather creating the datasets required to train them, an issue that is particularly pronounced in archaeology. Although the field is highly digitalised, it is less often transformed into tangible data.

That is, archaeological information is often stored in digital formats but is rarely structured in a way that makes it accessible or usable by machines, limiting its applicability to AI. Consequently, it has become increasingly vital to ensure that archaeological data is Findable, Accessible, Interoperable, and Reusable (FAIR), particularly in machine-actionable formats as defined by the FAIR Principles.

In September 2024, Managing Artificial Intelligence in Archaeology (MAIA) officially launched its network, bringing together digital and field archaeologists, experts in chronologies and diverse disciplines, data curators, computer scientists, museum professionals, science communicators, academics, and practitioners.

This COST Action aims to develop a shared understanding of AI applications in archaeology and in doing so MAIA brings together more than 255 members spread across 34 countries.

“The MAIA COST Action is a great opportunity for exchange among people across Europe and beyond who are engaging with the open challenges that AI poses daily to all disciplines, including archaeology. We are at a crucial moment of technological acceleration in many areas of archaeological practice. MAIA offers a unique opportunity to come together and collaborate as we try to understand which directions we would like — or need — to take in the coming years. The intersection of so many different professions and backgrounds offers the opportunity to bring together diverse perspectives and a wide range of expertise to tackle these challenges with fresh ideas.” Dr Gabriele Gattiglia, Chair of MAIA

MAIA tackles the complex and rapidly evolving relationship between archaeology and artificial intelligence, a field rich with potential, but not without its challenges. The network aims to assess the current state of the art in the application of AI in archaeology, identifying both its advancements and limitations. It seeks to create an international collaborative framework to explore the availability and characteristics of open archaeological datasets suitable for training AI systems, while also promoting the expansion of such resources.

Additionally, the network is focused on understanding the key archaeological research questions that AI could help address, with careful consideration given to resource demands, ethical implications, transparency regarding biases in data and algorithms, and the long-term sustainability of these technologies.

Finally, MAIA will strive to effectively communicate the challenges and opportunities arising from the intersection of AI and archaeology, sharing findings and insights with a broad and diverse public audience.

In May 2025, 70 participants from over 30 countries gathered in Seville for the Kick-off Meeting of MAIA hosted by the Facultad de Geografía e Historia, University of Seville.
The Seville meeting marked the official launch of these activities and laid the groundwork for MAIA’s future steps. Training schools and dedicated working group meetings will be launched and organised over the coming months. These will provide hands-on opportunities to map existing archaeological datasets and research AI applications worldwide, as well as to train researchers, particularly young researchers, in the creation of high-quality archaeological datasets suitable for training AI tools.

MAIA is now online with a dedicated website (www.maiacost.eu) and is active on social media platforms, including Instagram, Facebook, and LinkedIn, where specific campaigns and calls to action will be launched to raise awareness of MAIA’s goals and broaden participation globally.