Fascinating post by Scott Alexander: Gwern’s AI-Generated Poetry
GPT-2 is the language processing system that OpenAI announced a few weeks ago. They are keeping the full version secret, but have released a smaller prototype version. Gwern retrained it on the Gutenberg Poetry Corpus, a 117 MB collection of pre-1923 English poetry, to create a specialized poetry AI.
Extensive samples provided, with commentary:
This is all perfect iambic pentameter. I know AP English students who can’t write iambic pentameter as competently as this….It has more trouble with rhymes – my guess is a lot of the poetry it was trained on was blank verse. But when it decides it should be rhyming, it can keep it up for a little while. From its Elegy Written in a Country Churchyard fanfic …
Scott chose interesting examples to show how the AI can start off rhyming perfectly and then gradually the rhyming deteriorates; or how it can start off well and then deteriorate into complete gibberish.
Would you spot this as fake robot-generated poetry if no one tipped you off?
My heart, why come you here alone?
The wild thing of my heart is grown
To be a thing,
Fairy, and wild, and fair, and whole
Scott really, really liked this tidbit, and says:
That last line, with its ABAB structure, is actually brilliant even by the standards of human poets. “Fairy and wild and fair and whole”. I could say that all day. This has to be a coincidence. It’s not that good anywhere else. But even having something generally okay enough that it can occasionally blunder into something that good is great.
I have to admit, several bits of poetry worked out really well. I am now predicting that someone is soon going to start using this type of AI thing to generate lines and poems. I can see that working much better than using this sort of word-generator to write prose. Would it be cheating, to present poems as though you wrote them yourself, if whole quatrains or longer stanzas were generated in this way?
Click through if you have time. There’s a trick buried in the post, so do read the entire thing if you’ve got a minute.