Feb 7, 2024 09:15 PM
https://theconversation.com/how-ai-could...ion-222079
EXCERPTS: . . . Scientists looked through 3D X-ray images of the papyrus. They trained AI to read letters in the scrolls based on subtle changes left in structure of the papyrus by the ancient ink.
The AI was able to decipher and translate the ancient Greek word for “purple” on the scroll. Other ancient texts from the Indian sub-continent, for example, are in the Sanskrit language or Grantha script, which very few people can read and understand today.
Here again, AI systems are using language translation and text prediction to predict missing characters in old inscriptions.
[...] As with any other technology, AI comes with its challenges. The most prominent in the context of faith is likely to be misinformation and misrepresentation. We have already seen an example of this last year with the viral circulation of a deepfake image of the pope. In that instance, it was largely meant as a bit of fun.
However, with the possibility of the creation of deepfake images and voice cloning through AI costing as little as £10 (using existing online tools such as Eleven Labs and D-ID) we might not be too far from more worrying cases. Imagine extremist organisations using this technology to create propaganda videos with the faked words of religious leaders or to misrepresent religious symbols.
Similarly, language models trained on extremist views could create outputs at scale to spew venom against particular faith groups, leading to hate crimes and community tensions... (MORE - missing details)
EXCERPTS: . . . Scientists looked through 3D X-ray images of the papyrus. They trained AI to read letters in the scrolls based on subtle changes left in structure of the papyrus by the ancient ink.
The AI was able to decipher and translate the ancient Greek word for “purple” on the scroll. Other ancient texts from the Indian sub-continent, for example, are in the Sanskrit language or Grantha script, which very few people can read and understand today.
Here again, AI systems are using language translation and text prediction to predict missing characters in old inscriptions.
[...] As with any other technology, AI comes with its challenges. The most prominent in the context of faith is likely to be misinformation and misrepresentation. We have already seen an example of this last year with the viral circulation of a deepfake image of the pope. In that instance, it was largely meant as a bit of fun.
However, with the possibility of the creation of deepfake images and voice cloning through AI costing as little as £10 (using existing online tools such as Eleven Labs and D-ID) we might not be too far from more worrying cases. Imagine extremist organisations using this technology to create propaganda videos with the faked words of religious leaders or to misrepresent religious symbols.
Similarly, language models trained on extremist views could create outputs at scale to spew venom against particular faith groups, leading to hate crimes and community tensions... (MORE - missing details)