The artificial intelligence cooperative: READ-COOP, Transkribus, and the benefits of shared community infrastructure for automated text recognition

The resource titled "The artificial intelligence cooperative: READ-COOP, Transkribus, and the benefits of shared community infrastructure for automated text recognition" is published by Open Research Europe. The authors include Melissa Terras, Bettina Anzinger, Paul Gooding, Günter Mühlberger, Michaela Prien, Joe Nockels, C. Annemieke Romein, and Andy Stauder, among others.

READ-COOP is a platform cooperative that develops and hosts its own Artificial Intelligence (AI) and Machine Learning (ML) tools, specifically Automated Text Recognition (ATR) technologies such as Transkribus. Established on July 1, 2019, the cooperative aims to address the needs of its members while promoting responsible AI infrastructure.

As of October 2024, READ-COOP has 227 members from 30 countries and 235,000 registered user accounts. Transkribus has successfully processed 90 million digital images of historical texts, showcasing its effective application of AI in the cultural heritage sector. The cooperative approach emphasizes democratic decision-making and stakeholder involvement.

READ-COOP demonstrates that a cooperative model can sustain innovation in AI and ML infrastructures. The cooperative structure fosters community engagement and accountability while reinvesting earnings to improve services. The qualitative feedback from members indicates high satisfaction with the governance and utility of the AI infrastructure.

Transkribus, the ATR platform, has been a game-changer for historians, librarians, and linguists. It allows users to generate accurate transcriptions of historical documents without requiring technical know-how. The platform's first-mover advantage has led to widespread adoption, with transcriptions underpinning thousands of research projects.

Transkribus won the European Union's Horizon Impact Award in 2020, recognizing its significant contributions to the cultural heritage sector. This accolade highlights the importance of community-driven initiatives in fostering technological advancements.

The cooperative's development has been documented through qualitative questionnaires and reflection-in-action methods. This approach enables the cooperative to assess its governance structure, management, and community engagement, ensuring that it remains responsive to member needs.

The success of READ-COOP suggests that cooperative frameworks are well-suited for AI infrastructures initially funded by public grants. The authors recommend broader exploration of cooperative models for innovation in AI technologies, emphasizing their potential for responsible creation and governance.

READ-COOP serves as a model for how cooperatives can offer sustainable solutions in the AI and ML fields, particularly within the cultural heritage sector. Its unique structure supports not only technological innovation but also community engagement and equitable ownership, presenting a viable path for future AI initiatives across Europe.