Pang Ying:Hello, I’m Pang Ying. Welcome to the Episode 26 debate of Think Like a Debater.

This is our second futuristic debate about AI. The motion is“Do AI-created works have artistic value?”Zhan says no; I say yes. She plans to defeat me later with poetry and literature.

As the affirmative, I will open.

I have three arguments.

AI studies large bodies of art, finds patterns and techniques, then uses parameters and prompts to create new work unlike existing examples.

For example, commercial systems learn from popular music and generate new songs that audiences enjoy.

A 2018 Microsoft–Kyoto University poetry project learned from thousands of images paired with human descriptions or verse, studying imagery, rhythm, color, emotion, and metaphor before writing poems.

Humans selected stronger poems for a poetic Turing test. Both experts and laypeople often failed to distinguish AI poems from human ones.

These examples are already years old. Machine learning is advancing rapidly in poetry, lyrics, music, trailers, choreography, writing, and painting.

One striking example is the website“thispersondoesnotexist”，Each refresh produces an utterly realistic portrait of a person who never existed. It is generated bya generative adversarial networkand is commonly associated withdeepfake。

Everyone has creativity and something to express, but mastering painting, music, or poetry requires immense time and resources. What family can train a child in every art?

AI changes this. An app can help me create personal music by testing which outputs reflect my feelings. Recently, we made songs with musician Xu Fei: Zhan wrote lyrics, Xu composed and sang.