A man using a laptop in a trench discovering digital history revolution.

Generative AI and the (Tame) Digital History Revolution

Unlocking the Potential of Generative AI in Digital History Revolution

It’s probably hard to know what to do with ChatGPT. Since OpenAI released its groundbreaking chatbot last November, the media’s either heralded it as a revolutionary technology akin to the invention of fire or more often portrayed it as a dangerous plagiarism machine, just the latest technological snake-oil. Yet most people seem to know little about it other than that it allows students to plagiarize essays, that it can spit out good recipes, and write passable Shakespearian sonnets. You’ve also probably heard that it gets things wrong (so-called hallucinations). So aside from frustrating educators during marking season, what could ChatGPT possibly offer those interested in history?

Using Chat GPT in Historical Research

The short answer is that in the very near future, generative AI — of which ChatGPT is just one iteration — will transform how we research historical problems and analyze documents, less so how we write up the results. I like to say that I am an unlikely and inherently skeptical convert as I am a tenured history professor and I love writing — I’ve published many books and articles. But for the past eight months I’ve been working intently with generative AI and writing about my experiences. And the more I play around with it, the more curious and excited I get about its potential. And the things that fascinate me most about the technology have nothing to do with its famous writing abilities. To understand why I am fascinated by AI at the moment, we have to understand how generative AI works.

How Generative AI Works

ChatGPT is a web-based program — a chatbot—that lives on top of a Large Language Model (LLM) that was trained on a neural-network mimicking the structures of the human brain to predict the next word in a sequence of text. This is why some people have called LLMs stochastic parrots and accused them of being plagiarism machines that just reorganize and regurgitate information. But this is not entirely accurate. LLMs are revolutionary because when they were trained on trillions of words of text, they began to do something very unexpected: they started to exhibit capabilities that they had not been trained for like writing, translation, problem-solving, and coding.

Chat GPT's Impact

But it is also why they hallucinate and sometimes get things wrong: they are not recalling information verbatim from a database like a search engine might do and when they don’t know something they tend to provide a convincing but fictional answer. Yet this is getting to be less and less of a problem: as they models get larger and more powerful, they also start to hallucinate less frequently. There is a world of difference in both accuracy and capabilities between GPT-3.5 (the free version of ChatGPT) and GPT-4 (the paid version). If you’ve dismissed generative AI after trying the old ChatGPT last winter, try GPT4 before you give up on the technology.

Few people realize how strange and disruptive this all is: for all intents and purposes, LLMs actually appear to reason. While many in the academic community have dismissed this idea as a mirage, they’ve been equally unable to explain GPT4’s behaviour. While “reasoning” may not be the right word, what fascinates me is that when I ask GPT-4 to do something complex like parsing a series of difficult-to-follow instructions, it usually does just what I asked it to do. While it may be true that the “reasoning” it exhibits is illusory, if it produces an accurate translation of a document I am not sure it matters to most people whether the machine was actually thinking or just doing some really neat stuff that looks like thinking.

Beyond Writing Abilities

It is this ability to do complex things quickly that makes LLMs so revolutionary. For starters, LLMs are very accurate translators. They can also accurately summarize text, create graphs and charts from data, and even perform their own analysis of things like census records. The predictive capacity of LLMs is also powering new Optical Character Recognition technologies, like the one offered by Transkribus, which can accurately transcribe handwritten documents in a matter of minutes. As an historian, I am most interested in what happens when we bring all of these capabilities together. Very soon we will be able to deploy specially trained LLMs to process large amounts of data quickly, finding references to people and underlying patterns we would never be able to see. This will, in turn, open up new questions that would have been impossible to tackle in the past as well as new ways of doing history.

A Practical Example

Let me provide a concrete example. I am working on a biography of Alexander Henry the Elder, the fur trader. So much of that work involves going through countless ledgers looking not only for Henry’s name but also the names of his business partners in order to understand how his social and economic connections overlapped. Its tedious work to say the least. Now imagine that those records were all digitised. You could certainly perform a series of manual searches — or with a bit of python coding you could ask GPT-4 to map out Henry’s networks automatically.

That would certainly save time! But the amazing thing is that you could then ask GPT-4 to map out all of the business networks for all of the merchants in Montreal between 1760 and 1820, which it could do in a matter of a few minutes. To be clear: it would not write up an analysis of those networks, but you could get it to generate tables, charts, and graphs that you could then work from. Amazing, but let’s take that one step further: using the digitized records of the Programme de recherche en démographie historique (PRDH), it would be possible to map those economic networks onto social networks, looking at kinship networks and social mobility in detail. And if you wanted to visualize the results, you could even ask GPT4 to create a map to show how capital and human-relations evolved in tandem. Apply this to any subfield where we have a wealth of digitized data and let your imagination run wild.

AI's Role in Writing History

If you are worried that AI is going to start writing our history books for us, you can breathe a sigh of relief. It’s a struggle at the moment to get GPT4 to write even 750 words. That’s why the focus on AI’s writing abilities, rather than its analytical and pattern finding capabilities, is so misleading. What excites me about generative AI is that it makes a whole lot of things feasible that would have been impossible or too time-consuming to contemplate a few months ago. With AI, I think we are truly on the cusp of a new digital era in history. But it will be a much tamer revolution than most are predicting, akin to the way digital archival photography altered the research collection process over the past two decades. It will simplify things and speed them up.

Don't miss out on the latest news!
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

People also read

A desk with two monitors displaying the ai hallucinations.

Hallucinations in Generative AI Models

Discover the mathematical underpinnings behind AI hallucinations and the significance of latent spaces. Despite the ability to handle vast datasets efficiently, generative models sometimes produce "hallucinations," creative but inaccurate outputs.
Ivan Sysoev
Ivan Sysoev
September 12, 2023
3
min read
Generative AI
Digital Humanities
"The Ambassadors" (1533) by Holbein depicts two figures with scientific instruments, symbolizing Renaissance knowledge. A distorted skull at the bottom serves as a reminder of mortality.

Cliodynamics and Mathematical Models in History. Part 3

Artzrouni and Komlos's 1996 spatial model visually represents territorial dynamics in Europe from 500 to 1800 AD using a grid system. The model underscores the influence of a state's border position and suggests coastal countries form more predictably than inland ones. However, it highlights the limitations of solely using geopolitical mechanisms to predict empire dynamics. Turchin believes other factors, like Ibn Khaldun's concept of "asabiyyah" (collective solidarity), play a significant role in empire rise and fall.
Dr. Alexander Tsikhilov
Dr. Alexander Tsikhilov
August 17, 2023
3
min read
Mathematical Models in Historical Processes
Cliodynamics
Rubens' "A Hippopotamus and Crocodile Hunt" depicts a chaotic scene of men battling a fierce hippopotamus and crocodile amidst a turbulent waterscape.

Cliodynamics and Mathematical Models in History. Part 2

Peter Turchin utilizes the Lotka-Volterra (predator-prey) equation, originally designed to model population dynamics between predators and their prey, to understand the complexities of medieval agrarian states. These states, according to Turchin, can be viewed as oscillating systems influenced by variables like territory size and military success. Drawing from Randall Collins' geopolitical theory, Turchin identifies key parameters such as geopolitical resources, logistic loads, and peripheral position. The interplay of these variables results in non-linear relationships between territory size and rate of change, suggesting there's an equilibrium point beyond which territorial expansion becomes inefficient for the state.
Dr. Alexander Tsikhilov
Dr. Alexander Tsikhilov
August 10, 2023
4
min read
Cliodynamics
Mathematical Models in Historical Processes

FAQs

How can I contribute to or collaborate with the Historica project?
If you're interested in contributing to or collaborating with Historica, you can use the contact form on the Historica website to express your interest and detail how you would like to be involved. The Historica team will then be able to guide you through the process.
What role does Historica play in the promotion of culture?
Historica acts as a platform for promoting cultural objects and events by local communities. It presents these in great detail, from previously inaccessible perspectives, and in fresh contexts.
How does Historica support educational endeavors?
Historica serves as a powerful tool for research and education. It can be used in school curricula, scientific projects, educational software development, and the organization of educational events.
What benefits does Historica offer to local cultural entities and events?
Historica provides a global platform for local communities and cultural events to display their cultural artifacts and historical events. It offers detailed presentations from unique perspectives and in fresh contexts.
Can you give a brief overview of Historica?
Historica is an initiative that uses artificial intelligence to build a digital map of human history. It combines different data types to portray the progression of civilization from its inception to the present day.
What is the meaning of Historica's principles?
The principles of Historica represent its methodological, organizational, and technological foundations: Methodological principle of interdisciplinarity: This principle involves integrating knowledge from various fields to provide a comprehensive and scientifically grounded view of history. Organizational principle of decentralization: This principle encourages open collaboration from a global community, allowing everyone to contribute to the digital depiction of human history. Technological principle of reliance on AI: This principle focuses on extensively using AI to handle large data sets, reconcile different scientific domains, and continuously enrich the historical model.
Who are the intended users of Historica?
Historica is beneficial to a diverse range of users. In academia, it's valuable for educators, students, and policymakers. Culturally, it aids workers in museums, heritage conservation, tourism, and cultural event organization. For recreational purposes, it serves gamers, history enthusiasts, authors, and participants in historical reenactments.
How does Historica use artificial intelligence?
Historica uses AI to process and manage vast amounts of data from various scientific fields. This technology allows for the constant addition of new facts to the historical model and aids in resolving disagreements and contradictions in interpretation across different scientific fields.
Can anyone participate in the Historica project?
Yes, Historica encourages wide-ranging collaboration. Scholars, researchers, AI specialists, bloggers and all history enthusiasts are all welcome to contribute to the project.