Historians have embarked on a groundbreaking study using artificial intelligence (AI) to track the dissemination of astronomical knowledge across Europe during the early 16th century. This innovative research challenges the traditional notion of scientific revolutions being the product of "lone geniuses," revealing instead that astronomical thinking was a widespread educational foundation influencing varied disciplines from medicine to literature. The research findings were published on October 23 in the journal Science Advances.
Leading the study, Matteo Valleriani, a computational historian from the Max Planck Institute for the History of Science in Berlin, posits that the study maps the emergence of a proto-scientific community that traversed national boundaries in Europe. Through the implementation of AI, researchers were able to analyse an extensive collection of 359 astronomy textbooks, written between 1472 and 1650, which were used in teaching geocentric astronomy—an astronomical perspective positioning the Earth at the universe's centre.
These textbooks provided essential knowledge on the positions of stars, considered critical not only for astronomy but also for other fields like Greek and Latin poetry, indicating the integration of celestial understanding into the general educational framework of that era. Introductory courses in astronomy, where such textbooks were employed, were mandatory for students across Europe, reflecting the importance of this knowledge at the time.
The project involved digitising and meticulously analysing a dataset comprising 76,000 pages of historical text, images, and numerical tables. Each book presented unique challenges due to varying fonts, formats, and compositions that a single historian might struggle to assess in a lifetime. Using advanced machine learning techniques, the team identified 10,000 individual numerical tables within these texts. The AI was further trained to recognise and interpret the numerical content, a complex task due to the inconsistent table designs that appeared throughout the dataset.
Physicist and machine learning expert Klaus-Robert Müller from the Technical University of Berlin highlighted the difficulties in achieving uniformity in formatting among historical texts. AI's role was essentially to classify and compare these numerical tables, facilitating the detection of reprints, evolutionary ideas, or innovative methodologies in using astronomical data.
This exploration into early European scientific dissemination offers new insights into how knowledge was shared and evolved across borders, contradicting the conventional narrative that focuses on solitary figures such as Copernicus, Galileo, and Kepler. According to Jürgen Renn, a computational scientist from the Max Planck Institute of Geoanthropology, the findings add depth to our understanding of the scientific revolution as a collective rather than individual enterprise.
One significant outcome of the study is the dominance of textbooks originating from Wittenberg, Germany, in the 1530s. These texts, mimicked widely across Europe in larger markets like Paris and Venice, contributed to a more unified educational approach to astronomy. Valleriani remarks on the irony of Wittenberg, known for being the cradle of the Protestant Reformation, simultaneously fostering a science educational method that gained pervasive acceptance even as Europe was divided religiously and politically.
Additionally, AI-assisted research into historical maps traces the changes in perceptions of global climate zones, with earlier maps dividing the inhabited world into seven zones evolving into depictions with as many as 24 due to expanding European explorations.
Despite its contributions, the study acknowledges limitations—principally, the inherent incompleteness and bias possible in the historical record, which requires careful selection by historians. The researchers underscore that AI is an invaluable tool but cannot replace human interpretation and analysis.
As Renn remarks, AI's application in historical study is a remarkable tool for understanding history beyond isolated events, envisaging it as a continuum of human endeavours and intellectual development. This project demonstrates how future historians might integrate AI methodologies to explore history, utilising these modern innovations as complementary aids rather than replacements for scholarly insight.
Source: Noah Wire Services