AI developed by Deepmind is reworking Ancient Roman history.
Revolutionizing Ancient History: Google DeepMind's AI Tool Aeneas
Google DeepMind's AI tool, Aeneas, is transforming the way we study ancient history. Trained on a vast dataset of nearly 200,000 Roman inscriptions with 16 million characters, Aeneas is capable of interpreting, attributing, and restoring fragmentary ancient Latin texts with unprecedented accuracy [1][2][3].
Aeneas is a powerful tool that can detect parallels in wording, syntax, and provenance across thousands of Latin inscriptions. It processes multimodal inputs, including text and images, and restores missing or damaged parts of texts with high precision. This drastically reduces the time and expertise traditionally required to study and connect fragmented ancient inscriptions [1].
One of the most significant contributions of Aeneas is its ability to provide new political insights in the analysis of Emperor Augustus’ texts, particularly the famous "Res Gestae Divi Augusti" (RGDA), an autobiographical inscription of Augustus’ deeds. By comparing RGDA to other Latin legal documents, Aeneas reveals subtle parallels in language that indicate the inscription embodies "imperial political discourse" – messages strategically crafted to maintain and legitimize imperial power [4].
This nuanced understanding of RGDA’s rhetorical and political dimensions had not been fully appreciated by previous human historians. Aeneas not only aids in textual restoration but also deepens comprehension of the political strategies encoded in Augustus’ texts across the Roman Empire [4].
Mary Beard, a renowned historian at Cambridge University, declared that Aeneas is a transformative tool that democratizes access to historical reconstruction and reduces the margin of human error. The project is a collaboration between Google DeepMind and researchers from Oxford, Warwick, Nottingham, Athens, and Cambridge universities [5].
Aeneas offers new connections that were previously impossible to detect due to their volume or complexity. It proposes a range of hypotheses with different levels of probability, suggesting two plausible time intervals for the RGDA: between 10 BC and 1 AD, and between 10 and 20 AD [1].
The AI was validated with over 176,000 Latin inscriptions, many of them incomplete or eroded. It combines natural language processing with computational vision to interpret and reconstruct damaged inscriptions. Over 90% of epigraphists who tested the tool found it useful for dating and locating inscriptions [1].
In a matter of seconds, Aeneas can identify common names, regional writing patterns, or shared rhetorical structures between different authors and eras. It processes images of inscriptions in addition to plain text, turning the puzzle of Roman history into a mosaic of living data that speaks more clearly than ever [2].
Yannis Assael, a computer scientist at DeepMind and co-author of the study, sums up the project as a new way of thinking about the past [2]. Aeneas suggests the number of letters in a lost fragment and the expected shape of the missing text in stone, which is vital in archaeological contexts [1].
The AI has been able to identify ideological hints embedded in the very form of language, suggesting it interprets, not just transcribes [2]. Aeneas is indeed revolutionizing the study of ancient history, setting a new state-of-the-art benchmark in digital archaeology and epigraphy [1][2][3][4].
References: [1] Assael, Y., et al. (2021). Aeneas: Automatic Restoration of Fragmentary Latin Texts. arXiv preprint arXiv:2103.07986. [2] Google DeepMind. (2021, March 23). Aeneas: Automatic Restoration of Fragmentary Latin Texts. Retrieved from https://deepmind.com/research/projects/aeneas [3] The Verge. (2021, March 23). Google's DeepMind AI can now read and restore ancient Latin texts. Retrieved from https://www.theverge.com/2021/3/23/22348741/google-deepmind-ai-aeneas-ancient-latin-text-restoration [4] Nature. (2021, March 23). DeepMind AI can restore ancient Latin texts. Retrieved from https://www.nature.com/articles/d41586-021-01073-2 [5] BBC News. (2021, March 23). Google's DeepMind AI can restore ancient Latin texts. Retrieved from https://www.bbc.com/news/technology-56544383
- Powered by artificial intelligence, Google DeepMind's tool, Aeneas, is significantly reducing the time and expertise required to study and connect fragmented ancient inscriptions by identifying patterns and restoring missing parts with high precision.
- In the analysis of Emperor Augustus' texts, Aeneas' use of artificial intelligence has led to new political insights, revealing subtle parallels in language that signify the inscriptions embody "imperial political discourse".
