Journal of Information Sciences (Dec 2022)
I Trust Artificial Intelligence: The Issues, the Project, the Impact
Abstract
The project aims to use AI to establish and maintain the trustworthiness of digital records. To this purpose it uses the concepts and methods of archival diplomatics and Deep Learning (DL), a subfield of machine learning and AI. DL is achieved by creating artificial neural networks in multiple layers that learn from input data through supervised learning. The objectives are to improve existing tools and create new machine learning tools for various records needs, using image recognition, optical character recognition, text summarization and classification, and more. The decision-making responsibilities are shared among academic and non-academic partners. Several studies are conducted to achieve the projects objectives objectives. One study aims to identify Private Personal Information (PPI) in records and link the labeling to diplomatics labeling. another study focuses on using DL to recognize the identity attributes of digitized parchments from medieval times through a tool called PERGANET. This tool could have a wide range of applications, including recognizing individual writing styles, analyzing archival annotations, and making AI datasets publicly available. Another study focuses on documenting the AI process, drawing on multiple fields, including social sciences, explainable AI, and archival studies. The goal is to answer questions regarding accountability and transparency in the use of AI and provide guidelines for creating and preserving records of AI processes and their outcomes. The I Trust AI project aims to combine the strengths of different fields and produce practical results that benefit society and address the challenges posed by the rapidly advancing field of AI, and DL in particular, ensuring its responsible and ethical use while realizing its potential benefits in various fields such as archives, cultural heritage, and governance.
Keywords
