When Microsoft CEO Satya Nadella puts ChatGPT to work though, the verses aren’t faux Taylor Swift lyrics—they’re masterworks written by humans, sometimes centuries ago. Nadella, whose company has invested $13 billion in ChatGPT maker OpenAI, says his favorite use of the technology is to translate the great Persian poets and to better understand complex works of philosophy.

“I remember my father trying to read Heidegger in his forties and struggling with it, and I have attempted it a thousand times and failed,” Nadella said recently in an interview with podcast host Stephen Dubner of Freakonomics Radio.

“And [Heidegger’s] written this essay, somebody pointed me to; somebody said, ‘Oh, you got to read that because, after all, there’s a lot of talk about A.I. and what it means to humanity.’ And I said, ‘Let me read it.’ But I must say going and asking ChatGPT or Bing chat to summarize Heidegger is the best way to read Heidegger.”

Nadella is referring to Martin Heidegger, a German philosopher (and, it should be noted, a onetime member of the Nazi party), who wrote a famous essay on the role of technology for humanity. In oversimplified terms, Heidegger argues that technology spoils human existence. It’s safe to say, he probably wouldn’t like generative A.I.

ChatGPT isn’t just helpful in summarizing dense philosophy, either. It can translate poetry from one language to another, in what Nadella calls a “cross-lingual understanding” of poetry. 

Nadella is a known poetry aficionado. It’s why this is his favorite of all ChatGPT uses, he said in the interview.  

As a kid in Hyderabad, India, Nadella dreamed of reading Jalal al-Din Rumi, the famed Persian poet, in English and Urdu. ChatGPT finally gave him the long-sought translation, and, more important, the model translated it well. 

“In one shot, it does it. But the most interesting thing about it is that it captures the depth of poetry. So it somehow finds in that latent space meaning that’s beyond just the words and the translation. That I find is just phenomenal,” Nadella told Freakonomics Radio. 

While it may have worked well for Nadella, ChatGPT is still frequently hallucinating, or fabricating ideas and portraying them as factual. 

Nadella didn’t mention it, but hallucinations are certainly an obstacle to finding the meaning of life hidden in a translation of a Rumi poem. If you’re looking to show Walt Whitman to your non-English-speaking friend for the first time, maybe stick to an academic translation.