What Artificial Intelligence, that "externalized brain", brings into play in History

Every day it's harder for me to do my job, and at the same time, it's getting easier every day. It's easier to access books and articles online that would take me hours, kilometers, or millions of pesos to physically access. Today, with a few clicks of the right keys, I can access the daily news from the Gaceta de Buenos Ayres , nineteenth-century manuscript trials from Colombian archives, the first edition of Henri Grégoire's De la Littérature des Nègres , or the latest book on slavery published by Cambridge University Press . All in minutes, almost instantly, from my office at the university or from home. So easy.

If we're lucky, our published sources have been digitized with character recognition (OCR), and we can then search for keywords without the trouble of reading every page of decades, months, and days of press, or long, multi-volume treatises. Much of the historian's old work is reduced to just another series of clicks.

Until recently, we thought, "What won't be possible is to replace our ability to read manuscripts." And here we are, seeing the proliferation of projects and the reality of reading handwritten sources with and by AI. As an example, look no further than how the parish records compiled and made available on www.familysearch.org went from being transcribed in remote small towns in the U.S. by devotees of the Church of Jesus Christ of Latter-day Saints to being effectively interpreted by an AI.

Ah! We said, but only historians can write coherent and meaningful histories based on our readings, the sources we select, the historiographical conversation that our work entails. And yet, every day we see our students in classrooms—not colleagues in lectures and publications—ask an AI developed in Silicon Valley or China to point out the relevant bibliography on a topic, summarize it, suggest gaps, sources, and write the result in essays with the tone of Eric Hobsbawm or the prose of Mario Vargas Llosa . And even more, then transform them into a podcast or a meme .

And that's when the work becomes more difficult. We are working in the midst of a crisis that, as Gramsci suggested, is this living between the old that is dying and the new that is never born . A crisis concerning the meaning of historiographical work, the formation of new generations, and the social and political role of science and history.

To the challenges of the decontextualized hyperaccessibility of sources, the Head of the Department of History at the University of Pittsburgh, Lara Putnam , responded a few years ago by appealing to the irreplaceable nature of reading in context, in series, situated, indispensable for the complete understanding of those pieces of historicity that are the sources.

She was right. Not only because history is situated and requires meaningful contexts to be understood. Also because who writes history matters. I don't mean to say that only locals can think about and understand the history of the place where they were born or live. The past is an "other" territory, alien to everyone . Understanding it is an accessible task for those who commit themselves rigorously and sensitively, patiently and imaginatively to the community being studied, to its traditions, dynamics, and legacies. This political and even emotional dimension that allows us to understand historical subjects outside of algorithmic frameworks is what can be imitated and replicated by AI, but not generated (although the verb is so dear to that universe).

The work of contextualized and cultural interpretation that historians can perform is irreplaceable and is the result of training in the old tasks of reading, interpreting, searching for and processing sources, imagining new ones, and connecting them with experiences, collective imaginations, and cultural sites that are not necessarily relevant a priori . It is the transmission of these skills, this know-how, that is at risk. The predominance of a strongly visual, or audiovisual, culture takes time and patience away from reading. The temptation to grasp a summarized history, processed by AI and regurgitated for consumption and reproduction, is omnipresent.

It's a difficult time to write history, also because people are reading less and less . Faced with this, outreach strategies are multiplying, using other formats to bring the past closer to generations less willing to spend hours staring at letters that don't move on the page. It's possible that the plasticity of the human brain makes lasting and critical learning possible through listening, but it's also possible that the massively distributed ability to read and the capacity to critically interrogate narratives that may even be presented attractively but not based on verifiable facts are being lost.

On my more pessimistic days, I think it's possible that the generations that rely on the eternal availability of the externalized brain that AI implies will find themselves in a not-too-distant future having to pay exorbitant amounts to have that brain and will no longer have the habit of reading, nor access to material sources and ways of checking the past and the present outside of those private systems that are carrying out an original accumulation of knowledge arduously acquired by humanity and now privatized in a veiled but effective way.

In short, my job is getting harder every day because it's vital, but fewer people care about it. It's attacked by a government that rejects it for not commoditizable, while being swallowed up by AI companies to feed their bases and their business, which promotes a false sense of effortless understanding and discourages literacy training and critical thinking. A media expert told me that the solution to this will emerge from technology itself. Perhaps, hopefully. When students talk about ChatGPT as if it were a guru and complain about unions for organizing walkouts defending their right to public education, I allow myself to doubt. As I watch the new waves, I am already part of the sea.

Candioti is a historian, an Independent Researcher at Conicet (National Council of the Interior) and the Ravignani Institute, and an Associate Professor at the National University of Litoral. She is the author of A History of Black Emancipation (Siglo XXI).