Thanks to a recent interview that CEO Satya Nadella gave to Wired, we know now what you would ask an early version of OpenAI’s most powerful AI model, GPT-4, being CEO of Microsoft.
Nadella has a specific question that he uses to test the capabilities of a new AI model, and it revolves around the famous poet Jalaluddin Rumi.
“There is one query I always sort of use as a reference. Machine translation has been with us for a long time, and it’s achieved a lot of great benchmarks, but it doesn’t have the subtlety of capturing deep meaning in poetry,” Nadella told Wired.
“Growing up in Hyderabad, India, I’d dreamt about being able to read Persian poetry — in particular the work of Rumi, which has been translated into Urdu and then into English. GPT-4 did it, in one shot,” he added. “It was not just a machine translation, but something that preserved the sovereignty of poetry across two language boundaries. And that’s pretty cool.”
Rumi, who lived in the early 13th century, was a renowned Persian poet whose work is loved worldwide. Even today, people on social media and celebrities like Chris Martin, Madonna, and Tilda Swinton admire his poetry.
However, translating poetry can be challenging. As Robert Frost once said, “poetry is what gets lost in translation.” And when you translate a work that has already been translated before, the difficulty increases.
There is a risk of losing the essence of the original work, similar to what happens in the game of telephone. While technology like Google Translate is impressive, human translators specialize in preserving the subtleties and nuances of the original text as much as possible.
For Nadella, a true test for an AI model is whether it can go beyond mere “machine translation” and maintain the “sovereignty” of Rumi’s work. He explained to Wired that witnessing GPT-4 achieve this was an incredibly astonishing experience for him and called it “a mind-blowing experience.”