| I wonder how AI (ChatGPT) would think differently if its base language was something other than English. I asked it yesterday about Ancient Egyptian interested also to see how math (Greek/Egyptian) would be represented, but
ChatGPT on the language itself: "To provide an example, let's consider a simple statement in English and explore how Ancient Egyptian could potentially convey additional information or nuances: English statement: "The sun rises." Equivalent statement in Ancient Egyptian (not a direct translation): "Ra, the life-giving sun, majestically ascends in the sky, illuminating the world and ushering in a new day." In this example, the English statement conveys the basic idea that the sun rises. However, the Ancient Egyptian statement incorporates additional cultural and religious associations. It mentions Ra, the ancient Egyptian sun god, attributing the sun's rising to Ra's life-giving aspect. It also includes descriptions such as the sun's majestic ascension, its role in illuminating the world, and its symbolic significance of ushering in a new day. By incorporating specific cultural references and poetic descriptions, the Ancient Egyptian statement provides a more elaborate and nuanced understanding of the sun's rising, capturing the broader context and symbolic layers associated with this natural phenomenon within the ancient Egyptian worldview. Please note that the above example is a hypothetical representation and not a direct translation, as the complete nuances and intricacies of the Ancient Egyptian language and culture cannot be fully captured in English. It serves to illustrate the potential for additional information and cultural richness that can be conveyed in Ancient Egyptian beyond the basic content of an English statement." |
Do you really think that Ancient Egyptians didn't have a simple declarative statement that "the sun rises"?
Don't fall into exoticism. Ancient Egyptians were human beings speaking language for the purposes of communication in their day-to-day life, just like we do.
That being said, it definitely could be interesting to see how LLMs would work when trained on different language corpuses and other cultural contexts.