On History, Historical Materialism and Artificial Intelligence

What is the role of a historian? Historians research, analyse and interpret the past, extracting meaning and establishing patterns from evidence (Anderson, 2004: 81-82). This evidence is often hard to find, it's located in the ground or within the archival depths. Once found, they work with primary sources; eavesdropping on the past to construct arguments about what happened and weighing up evidence based on scraps of information (Galgano et al., 2008: 1-5, Devon et al., 2012: 122).

This standard historical method, however, is shifting due to the increasing availability of computational methods. Adam Crymble, in his 2021 book Technology and the Historian, stresses that: ‘As we [historians] increasingly work with computers, computers will increasingly exert their influence on our intellectual agenda, and so we must understand both their strengths and the limits they impose on us’. After ChatGPT’s launch in November 2022, and the resulting AI boom, this intellectual exertion appeared all the more visible. Anxieties around no-code approaches, vibe coding, labour displacement, in the case of graphic design and manual labour especially, and general cognitive offloading are now intermingling with the recognition that public understanding around AI is limited (Cave et al., 2018). With technologies' inner-workings often obscured, we are increasingly relying on our personal reception to AI and public discourse, which tends toward the hyperbolic (Gooding, Terras and Warwick, 2013). It is therefore incumbent on historians to think about technology, its development and how it can be appropriately used for reconstructing the past.

For those in the Digital Humanities, this is well-trodden ground. The expectation that historians, or humanities professionals in general, should code themselves re-ignites the old ‘hacks versus yacks’ debates of the 2010s. These debates recognised that technical skills involved a steep learning curve; however, they also contended with being at the whims of those more technical. Among historians in particular, those more hesitant to digital methods cited a loss of contextual knowledge in using such approaches: that finer information is missed, or changed, when relying on rules-based technologies (Zaagsma, 2019). In other cases, historians suggested that the material aspects of the past cannot be fully expressed through digital surrogates (Jeurgens, 2013). More recently, historians have expressed angst around the environmental sustainability of using AI, with server farms changing microclimates as close as Ireland; while water usage has skyrocketed at a time of global drought (McCabe, 2026). These issues resonate, even in pro-digital circles, yet - as Ortolja-Baird (2026) suggests, little else is known of how historians are using, critiquing or receiving AI. 

*** 

What all historians do agree on, however, is that the fundamental concept of history cannot be separated from its production: that control over people, labour and industry marks the stories we tell ourselves and each other. Here, we return to Crymble’s argument that computers and more recent AI technologies are part of that historiographical production. You cannot simply pick up history and use it; in the way you can with a GenAI model. “We need history, but we need it differently from the spoiled lazy-bones in the garden of knowledge”, as Nietzche states, with the practice of telling history requiring effort, interpretation and insight. Otherwise, those bones are made unambiguous, as if just left there and strategically placed for all to see by certain actors. For historical materialists, those actors are hardly disguised; they are the ones in control of the means of production. 

Benjamin offers an alternative to the lazy-bones use of history, that the past is under constant construction and fulfilled only by the here-and-now [Jetztzeit]. It is situated but only for a moment. In those moments, passing by us every second, we have an opportune moment to understand ourselves and our past, albeit within an “arena in which the ruling classes are in control”. 

How, then, does the digital come into play? 

*** 

The easier reading would be that large-scale AI companies are antithetical to knowledge custodianship, instead being driven by extraction and profit motives (Gooding, 2025). In this view, digital technology is merely an extension of our current capitalistic moment and its development falls neatly within a historical materialist’s view of the past. These drivers govern how chat-based interfaces are developed, with their constant want to be prompted - would you like me to do that for you? I can provide research-grade code for your entire pipeline. They also dictate the webcrawls models rely on, extracting whatever data they can find, whether dross or thought-out prose. Like a shark that can’t stop swimming through fear of suffocation, water passing over gills without interruption, they need constant data.

Digital scholars are attempting to change this moment for the better, opting for co-operative business models that place users directly within AI development and associated decision-making (Terras et al., 2025) and constructing knowledge elicitation frameworks to circumvent the fact that models are outpacing available training data (Alteras, 2026). However, the actions of certain AI companies are doing little to dispel the argument that their aims are extractive and purely capitalistic. The historical materialist is largely proven correct. This is usually where public discourse, and indeed most academic discussion, ends. 

*** 

Walking tour through Trondheim, Norway, using Augmented Reality to create distinct digital experiences from physical space raising questions of historical authenticity. In this case, 3D models were supplanted onto sites around the city, as part of the European Association of Research Libraries (LIBER) Annual Event and its workshop: Generative Humans: Mediating Automation & AI Through Impactful Engagement and Experience with Cultural Collections. Own image.

The historical materialist states that the production and distribution of goods underpins current and past society. Is the digital part of that process as suggested above, or does it offer up another superstructure? Put simply, does it bridge our current capitalistic era or present a newer murkier, but hoving into view, societal stage? 

Never before has history been told through the very same materials as used in its production. Mill owners, whom we imagine to be like John Thornthon in North and South (1855) - ordering children under looms to collect loose threads, were not using the cloth they produced to analyse history. Nowadays, certainly in the Global North, we are engaged in transmitting, questioning, encountering, interpreting and problematising history through a digital lens that itself is materialistic. The ink used by medieval chroniclers, or the pencil lead used by 20th century diarists, had a market, came from somewhere and was part of trade and business; they did not influence how the historian retold the past. With the digital, our very method of eavesdropping on the past has altogether shifted. The digital has its own epistemological concerns … 

The historical materialist stresses that the subject of history itself is in the battling, oppressed class, whereby each working-class is seen as the saviour of its future generation. Similarly, commercial GenAI companies are using this perpetual future tense. Their tools are marketed as alleviating manual labour, bridging previous skill gaps and opening up greater knowledge, intellectual power and revenue. Our lifeworld, which revolves around the use of technology, is now shrouded in the language of historical materialists, repurposed to evoke a similar ideal of emancipation but with very different results. With the proliferation of digital flotsam and jetsam - posts, reels, AI slop, historical authenticity [‘aura’ as Benjamin coins it] becomes harder to ascertain. 

This is not because AI tools throw up errors or hallucinate with increasing confidence, but because, in the case of commercial GenAI companies, models are very much the master’s tools. History is there, being constructed constantly, and we are losing track of how to talk about it … 

References 

Alteras, N. (2026) “Teaching Machines to Learn Like Humans: The Quest for Data-Efficient AI”, Paper presented at Centre for Machine Intelligence Townhall, University of Sheffield, March 11, 2026.

Anderson, I. (2004) “Are you being Served? Historians and the Search for Primary Sources”, Archivaria. 58: 81-129. Available at: https://archivaria.ca/index.php/archivaria/article/view/12479.

Benjamin, W. (1940) On the Concept of History, trans. Dennis Redmond, 2005. Available at: https://www.marxists.org/reference/archive/benjamin/1940/history.htm

British Museum (2026) “Endangered Material Knowledge Programme”. Available at: https://www.britishmuseum.org/our-work/departments/africa-oceania-and-americas/endangered-material-knowledge-programme

Cave, S. et al. (2018) “Portrayals and perceptions of AI and why they matter”, The Royal Society. Available at: https://doi.org/10.17863/CAM.34502.

Devon, E., MacDougall, R., Turkel, W.J. (2012) “New Old Things: Fabrication, Physical Computing, and Experiment in Historical Practice”, Canadian Journal of Communication, 37(1):121-128. 10.22230/cjc.2012v37n1a2506

Galgano, M.J., Arndt, J.C., Hyser, R.M. (2008) Doing History: Research and Writing in the Digital Age. Boston: Thomson Wadsworth.

Gooding, P. (2025) ‘AI in the National Library: Thinking Fast and Slow’, Artificial Intelligence Meets Cultural Heritage. 39th Annual General Meeting of the Conference of European National Librarians (CENL), Edinburgh.

Gooding, P., Terras, M. and Warwick, C. (2013) “The Myth of the New: Mass Digitization, Distant Reading, and the Future of the Book”, Literary and Linguistic Computing, 28(4): 629–639.

Hays, K. (2026, July 11) “Meta pulls new AI image feature after days of backlash”, BBC News. Available at: https://www.bbc.co.uk/news/articles/c2dy6e8klw0o

Jeurgens, C. (2013) “The Scent of the Digital Archive: Dilemmas with Archive Digitisation”, Bijdragen en mededelingen betreffende de geschiedenis der Nederlanden, 128(4): 30-54. 0.18352/bmgn-lchr.9348.

McCabe, B. (2026) “Digital practices in two regional theatres and arts centres, at the beginning of their sustainability journey”. Paper presented at UK-IE Digital Humanities Association, University of Southampton, June 15, 2026.

Neitzsche, F. (1874) On the Use and Abuse of History of Life. Available at: https://la.utexas.edu/users/hcleaver/330T/350kPEENietzscheAbuseTableAll.pdf

Ortolja-Baird, A. (2026) Survey on AI use by historians. Available at: https://docs.google.com/document/d/181ZzzQPJI_b8xAjcv5MZGUNlKkfV4o6QX_OAVZCO5WA/edit?tab=t.0.

Terras, M. et al. (2025) “The artificial intelligence cooperative: READ-COOP, Transkribus, and the benefits of shared community infrastructure for automated text recognition”, Open Research Europe. Available at: https://open-research-europe.ec.europa.eu/articles/5-16.

Wilson, T.D (2006) “On user studies and information needs”, Journal of Documentation, 62(6):658-670. 10.1108/eb026702

Zaagsma, G. (2019) “Digital History and the Politics of Digitization”, Digital Scholarship in the Humanities, 38(2): 830-851. Available at: https://academic.oup.com/dsh/advance-article/doi/10.1093/llc/fqac050/6702047