From the video production industry to the medical field, artificial intelligence is driving new discoveries. However, a new field that AI is currently entering is the study of ancient secrets.
The ZDNET website reports that Italian researchers at the University of Pisa have successfully employed artificial intelligence to decode a papyrus manuscript discovered in Herculaneum, a town close to Pompeii that was also destroyed in the year 79 after the eruption of Mount Vesuvius.
The manuscript is one of 1,800 that were saved from being buried in the muck and ash left by Mount Vesuvius at the Villa Papyri, which was once owned by Julius Caesar's father-in-law.
The scrolls must be interpreted using a hands-free scanning method since they are too delicate to handle due to their charring.
The researchers employed optical coherence tomography (OCT) and infrared hyperspectral imaging (IRSI) to view the burned papyrus, according to the Italian news agency ANSA.
The researchers found Plato's ultimate resting place, a garden setting on the site of the Platonic Academy in Athens, by recognizing and translating 1,000 words, or around 30% of the text.
Additionally, the text shows that Plato was not sold into slavery in 387 BC as historians had previously believed, but rather around 404 or 399 BC.
The finding validates the technology's ability to advance our understanding of this era's historical details and most notable individuals.
Following AI advances in February of last year, the Vesuvius Challenge—a global competition to decode the entire collection of the Herculaneum Scrolls—was started in March 2023. fresh insights into history.
Without taking the chance of actually damaging the scrolls, the project's fundamental technology decodes the writing on them using a combination of CT scanning and machine learning.
One of the challenge's organizers, Brent Sales, a researcher at the University of Kentucky, divided the procedure into three stages: segmentation, ink detection, and scanning. After segmenting the scrolls into individual pages and taking accurate cross-sectional scans of their interiors, the researchers use machine learning to decipher the writing on each page.
It is challenging to discern the scrolls by computer because they are basically black on black due to the volcanic eruption's charring of the ink and scrolls.
Sales collaborated with researcher Stephen Parsons to create a machine learning algorithm to read carbon writing. The group then produced Volume Cartographer, an open-source text-interpreting software, according to ZNet.
