An Emergence from Winter or Summer may be upon us
It is a bit of nail-biter writing anything about AI with a longish publication timeline. I submitted the following piece as an afterword for “Navigating Artificial Intelligence for Cultural Heritage Organizations” (UCL Press June 2025) last Fall. Despite a few months passing, and a few months to go before broader release of the volume I think the afterword holds up well!
It seems that an AI summer may be upon us. This seeming-summer is experienced variably, at turns engendering a sense of possibility and a sense of disquiet under the glare of for-profit AI marketing, data dependencies derived from us but refused from us (a kind of data gaslighting), and rarefied computing infrastructure distilled to freemium service models that work to incentivize transition to paid services controlled by far too few actors operating in to be determined regulatory environments. Like with anything worth doing, there is much at stake in determining how cultural heritage organizations will make the best use of AI. Navigating Artificial Intelligence for Cultural Heritage Organizations is an important asset to cultural heritage organization strategy and practice insofar as it introduces cultural heritage specific use cases as well as theoretical explorations that should help cultivate increased confidence. Whether explicitly or implicitly stated, the work of this volume is anchored by the long history of cultural heritage organizations using computational means to meet the needs of the communities they aim to serve - whether that be realizing the potential of the machine readable catalog, wide spread internet access, digitization, or continuous efforts to evolve library ability to foster literacies for a diverse range of communities. Time is a circle, not an arrow as some would have us believe. There is confidence to be gained in the knowledge that cultural heritage organizations and allied disciplinary researchers have continuously engaged with the potential of technology. We are not disrupted, we stand ready to make the best of technology as we have always sought to.
“All of this has happened before, and all of this will happen again.”
‘The hand of God’, S01E10, Battlestar Galactica (2005)
45 years ago, Linda Smith, submitted her dissertation, Selected artificial intelligence techniques in information retrieval systems research. Smith's dissertation, "reports on the results of research which has explored possible contributions of artificial intelligence (AI) to the design of information retrieval systems." Smith goes on to engage with aspects of pattern recognition, representation, problem solving, learning, and query formulation as problem reduction. Was this work prescient or practical? Both characterizations are equally salutary and demonstrate that engagement with the potential of AI is longstanding in library and information science. Smith would go on to mentor multiple generations of cultural heritage professionals and scholars, leading the Graduate School of Information Science at the University of Illinois at Urbana Champaign to adapt to and forecast fundamental changes in the global information ecosystem.
24 years ago, John Unsworth delivered a talk at King’s College titled, Scholarly Primitives: What methods do humanities researchers have in common, and how might our tools reflect this? In the talk Unsworth proposed 7 activities (primitives), common to scholarly work regardless of discipline, that computational tools should support: discovering, annotating, comparing, referring, sampling, illustrating, and representing. Unsworth proceeded to represent these primitives as axiomatic or self-evident bases for informing the design of computational tools. His work continues to be instructive for the present moment insofar as it provides the basis for a heuristic that helps articulate well-scoped, self-evidently valuable activities that artificial intelligence should support rather than getting caught in a collectively regrettable, technology-tail wagging the dog moment.
2 years ago, Emily Bender said the following while debating the nature of large language models (LLMs) at a conference, “I feel like there’s too much effort trying to create autonomous machines … rather than trying to create machines that are useful tools for humans.” 8 years prior to that Trevor Owens, delivered a talk titled, Mecha-Archivists: Envisioning the Role of Software in the Future of Archives that contained a similar sentiment:
My vision for the future of the archivist using digital tools is less Borg and more Exo-suit … The idea of mecha or exo-suits, illustrates a vision of technology that extends the capabilities of it’s user … We need tools that let us quickly work across massive amounts of items and objects by extending and amplifying the seasoned judgment, ethics, wisdom, and expertise of the archivist-in-the-machine.
Bender and Owens do important work here, cutting through the chaff to focus our interaction with AI as user driven and utilitarian in nature. No need to anthropomorphize technology here — just a powerful set of tools to support core activities that extend our imperfect, eminently human judgment.
With Smith’s orientation to information retrieval, Unsworth’s heuristic, and Bender and Owens’ user-driven framing in mind, librarians and archivists should be able to engage with the potential of AI clear-eyed and confident. We can have a relationship with these tools akin to the mutualistic relationships that we have with any tool — think of the virtuous loop between you and a bicycle, a pan, or camera. You acquire them for a certain purpose and you refine your purpose as you experience the world through them — investing in better brakes, cast iron seasoning techniques, and lenses. It’s a sort of symbiotic relationship between human and tool that changes usefully over time.
Much has been made in this piece of the power of long histories to draw upon, yet that is not to suggest that those histories are perfect. The imperfect aspects of cultural heritage organization histories should help us equally with the work that lies ahead. Well-documented harms inflicted by cultural heritage organizations are a painful yet instructive motivation. With the help of leaders like Lae’l Hughes-Watkins we work to deal with our past and that dealing helps keep us vigilant about how we move and ultimately makes us more responsive to the people we aspire to support in the present. It keeps us focused in the ways that matter most. It helps us do the right thing. We can do the right thing. We will do the right thing.