O CHAM irá receber, no dia 21 de Maio, um evento sobre arqueologia e inteligência artificial no âmbito da COST Action MAiA - Managing Artificial Intelligence in Archaeology.

As part of the COST Action CA23141 “Managing Artificial Intelligence in Archaeology” (MAiA), this meeting convenes a diverse audience – archaeologists, researchers, AI and IT specialists, publishers, GLAM sector professionals, policy makers, data managers, students, and communities affected – to chart the future of AI in archaeology. The event will tackle pressing challenges in building AI training datasets and utilising comparative collections, explore new frontiers in archaeological research enabled by AI, and engage in crucial ethical discussions. Its primary objective is to foster a broad exchange of views, empowering all stakeholders to actively participate in defining AI's role in understanding the past.

Despite popular perception, one of the most challenging aspects of AI is not the development of appropriate algorithms, but the creation of the datasets necessary to train them – and archaeology represents an extreme example of this problem. While archaeology is increasingly digital, it is rarely datafied, usable and machine-actionable, thereby limiting the effective application of AI. Even where digital collections exist, significant barriers to access remain, including restrictive copyright regulations. Ensuring that archaeological data are FAIR (Findable, Accessible, Interoperable, and Reusable; Wilkinson et al, 2016), CARE (Collective Benefit, Authority to Control, Responsibility, and Ethics; Carroll et al., 2020), and machine-actionable (as outlined in, though not limited to, the FAIR Principles) has therefore become paramount (MAiA, 2024).

The aim of this workshop is to bring together stakeholders to discuss how to overcome future challenges related to the use of archaeological comparative collections for AI applications, and to foster competencies in building training datasets for AI applications. Main topics are:

- Limits and problems of AI applications in the creation and reuse of comparative digital collections.

- Currently available comparative collections – attention will be paid to the legal frameworks around copyright, intellectual property rights (IPR), open licensing, data and metadata availability/access (Open Science, FAIR, CARE), ethics, and long-term sustainability.

- Prepare next steps to develop best practices for digitising training data for AI applications, and guidelines for creating or augmenting this data in ways that can be made actionable by machines.

- Increase the number of digital comparative collections available online, along with the metadata for FAIR machine-actionability necessary for AI use.

Cartaz (.jpg)

Organização

COST Action MAiA - Managing Artificial Intelligence in Archaeology (União Europeia)

CRIA / IN2PAST / NOVA FCSH

FLUP / U.PORTO

Centre for Digital Culture and Innovation

ANTIGA MENTE

FCT - Fundação para a Ciência e a Tecnologia