Future-proofing archives: Embracing AI for diverse user needs

In an era marked by rapid digital transformation, the role of archives is evolving dramatically. With an exponential increase in born-digital materials, archives face unprecedented challenges and opportunities in preservation, accessibility, and user engagement. Artificial Intelligence (AI) is at the forefront of this shift, promising innovative solutions to complex archival challenges and enhancing user experiences. However, integrating AI into archival practices demands a nuanced understanding of diverse methodological requirements and ethical considerations.

The recent study titled "Conceptualising methodological diversity among born-digital users: insights from the garbage can model," authored by Adam Nix, Stephanie Decker, and David A. Kirsch, and published in the renowned journal AI & SOCIETY, critically explores these dynamics. The paper provides a comprehensive analysis of the potential and challenges of adopting AI within archival systems, highlighting the distinct methodological needs of different user groups.

Unpacking the methodological diversity

The study brings forward the importance of recognizing diverse user groups interacting with born-digital archives. By employing the Garbage Can Model (GCM) - originally developed to understand organizational decision-making - this research identifies how AI integration within archives unfolds organically rather than systematically. Users approach archives differently based on their methodological needs, raising the risk of one-size-fits-all technological solutions which might inadequately serve distinct research approaches. The authors introduce a typology of born-digital archival users: aggregators, synthesisers, fact-finders, and narrators, each demonstrating unique methods and motivations in engaging with digital materials.

Understanding users and their differing methodological needs is central to the effective deployment of AI solutions in archives. Aggregators seek large-scale quantitative datasets for statistical analysis, emphasizing completeness and representativeness of data. Synthesisers, however, prefer thematic exploration and contextual richness, leveraging archives primarily for interpretative research. Fact-finders have sharply defined informational needs, valuing precise and exhaustive searches critical for tasks such as legal investigations or journalistic inquiries. Narrators use archives narratively, tracing detailed historical stories or genealogy, where depth and historical detail are essential.

The Garbage Can Model provides a compelling theoretical framework to explain how AI tools in archives emerge from the coincidental intersection of technology, user demands, and organizational solutions. Rather than methodically planned implementations, archival practices frequently evolve through ad hoc decisions influenced by available technologies and immediate user demands. This non-linear and iterative process highlights the challenge in systematically predicting or addressing user needs, underscoring the necessity of inclusive dialogue among technologists, archivists, and diverse user communities.

Ethical and regulatory complexities

Introducing AI into archival practices brings forth significant ethical and regulatory challenges. The study emphasizes issues surrounding data privacy, intellectual property rights, authenticity, transparency, and potential biases within AI systems. Archivists and technology developers must grapple with these ethical dimensions, such as ensuring user privacy, addressing biases inherent in AI algorithms, and navigating complex data ownership and protection laws. Transparency and accountability are paramount, requiring comprehensive ethical guidelines and regulatory oversight to build trust among all stakeholders. Addressing concerns about algorithmic transparency, data bias, and ethical use of archival data becomes critical, demanding carefully crafted governance frameworks that foster trust and accountability.

Moreover, ethical dilemmas extend to decisions about what materials are preserved and how access is managed, balancing the public's right to know with individual privacy concerns. The increased capability of AI to process vast amounts of data swiftly further compounds these concerns, potentially amplifying risks of misuse or misinterpretation. Hence, proactive strategies for ethical risk management and informed consent must be integral to archival practices as they adopt AI technologies.

Future directions: Collaborative innovation

Despite challenges, AI tools offer immense potential to transform archival practices, making vast digital collections more accessible and useful across various user types. AI can enhance discovery capabilities, automate metadata generation, improve search functionalities, and facilitate complex analytical tasks previously impractical or overly time-consuming. Such tools can also enable archives to serve underrepresented communities better, ensuring more equitable access by contextualizing and presenting archival material relevantly and responsibly.

The authors advocate for a collaborative approach in developing AI-driven archival tools, emphasizing the need for cross-disciplinary communication between archives, AI developers, and user communities. Future progress hinges on better integrating user perspectives into technology development, anticipating methodological needs, and addressing ethical concerns proactively. By facilitating inclusive dialogue and participatory design, the archival community can leverage AI's capabilities effectively, ultimately enhancing access and preserving digital heritage meaningfully and ethically.