AI tech exposes various possibilities: neural networks learn to paint, write music and texts. Artificial intelligence is currently used by historians: they decipher ancient writings and create a single base of historical data.
Let us see how AI is used to learn the past and how it can help discover new historical facts.
AI’s importance for history as a science
Artificial intelligence is currently just starting penetrating the sphere of history. There are not so many examples but the specialists have already singled out its potential.
For example, in 2012, researches from the Swiss Federal Institute of Technology in Lausanne co-operated with national libraries and museums of many countries and created the Time Machine Project. They plan to develop a digital model of European towns of the past based on archive documents and museum pieces.
The experts are now working on the platform that reflects 2000 years of European history. It would contain cultural, social, and geographic changes. What is more, the will develop the technology to analyze the accumulated data at a large scale.
To obtain effective analysis, the data has to be accumulated in one place, so the scientists plan to combine archives of European towns into a single database.
This solution would make it easier for historians to compare the data, analyze changes, and obtain new information from ancient texts.
AI-driven data analysis would allow to avoid subjectiveness. The experts are frequently influenced by the time period they are living in, which affects their view on historical facts.
Besides, authors of historical books can express a biased opinion on events omitting some information. When the data is collected by algorithms, structured and sifted through artificial intelligence, the events will be presented as they are.
Current use cases of AI for historical purposes
AI deciphers the Vatican Archives
Artificial intelligence and a face recognition system has already been used to decipher the Vatican Archives, an ancient trove of writings with documents. Project In Codice Ratio used artificial intelligence and optical character recognition (OCR) in order to digitalize the texts.
OCR only works on typeset characters. The technology sees words as images of separate letters, compares images with learned letters, and turns them into a source code.
The scientists have developed the analogical technology that breaks words into characters: the algorithm learns vertical and horizontal bars composing letters. This AI has been trained by Italian students who assessed the algorithm performance and created a special storage bank.
Thus, AI learned 22 letters of Medieval Latin and was trained to calculate the probability of their combinations.
As a result, the algorithm managed to decipher several documents from the Achieve, including letters to the European kings and orders. AI correctly recognized two thirds of the words deciphering about 96% of letters. As reported by one of project heads, even such inaccurate information was enough.
AI sifts through old newspapers
Researchers from University of Bristol used AI to analyze British newspapers issued within the last 150 years. Their goal was to find out whether this information may help detect historical and cultural changes in the society.
The algorithms analyzed 35 million newspapers. The AI not only pinpointed the milestones but also the changes that took part within the long period of time such as the introduction of a new technology or political ideas.
In such a way, automated content analysis detected key events such as wars, epidemics, and coronations as well as recognized the connection between people and geolocations.
AI’s prospects for historical studies
Artificial intelligence is very prospective from the point of view of historical studies: math processes would bring objectiveness while AI vision would help read ancient texts.
By combining available historical data with technology’s ability to store and sift through data, experts would obtain new information from the existing sources.
AI experts will discuss the prospects of the technology at AI Conference in Moscow.