Singapore stone mystery nears resolution with AI breakthrough

The enigmatic sandstone was discovered in 1819 in Singapore. Workers clearing trees from the southeastern side of the Singapore River found it entirely by accident. The rock measured a few metres and was covered with about 50 lines of text.

Despite a petition filed by Lieutenant Colonel James Low, the mysterious rock was blown up in 1843 to widen the mouth of the Singapore River.

Low went to the river, where he gathered the remnants of the sandstone with visible inscriptions. He sent them to the Royal Asiatic Society Museum in Calcutta (now the Indian Museum), where attempts were made over the following years to decipher the unknown script.

For the next few decades, the mysterious script could not be deciphered. Scientists determined that the writer was most likely someone from Sumatra who used the Old Javanese language.

This is still not enough to determine what was written on the stone—which, incidentally, the Singapore National Museum, where the mysterious object is currently located, refers to as one of the twelve "national treasures."

Many years have passed since the discovery of the Singapore Stone, and the only effective research was conducted by the Dutch epigraphist Johan Hendrik Caspar Kern, who managed to decipher a few words but could not identify the language in which they were written. He estimated that the inscriptions on the sandstone might have been made in 1230.

Another Dutchman, N.J. Krom, concluded that the script on the stone is similar to that used in Majapahit, a Hindu-Buddhist kingdom on Java that existed from 1293 to 1550. He also stated that the script might date from before 1360.

Finally, one of archaeology's greatest enigmas may be solved with the help of artificial intelligence. Still, it will be an exceptionally difficult task because, so far, the language found on the Singapore Stone has not been identified anywhere.

Artificial intelligence will primarily use the only available fragment of the rock but also drawings of the entire sandstone from 1837. For crypto linguistics, collecting as much material with the unidentified language as possible is crucial, making it easier to recognise.

A team led by Francesco Cavallaro from Nanyang University in Singapore oversees the tool to decipher the script on the Singapore Stone. According to announcements, AI will be more effective than humans as it can bypass limitations associated with cognitive errors. The tool will not only learn previously unknown characters but also estimate what might have been on the missing sandstone fragments.