Over at SlateStarCodex.com, Scott Alexander reviews poetry. For example:
Thou know’st how Menoetiades the swift
Was dragged, of Hector and the fierce compeers
And Phrygian warriors. So, we will dispatch
Your bodies, then, yourselves to burn the ships
In sacrifice; with torches and with bells
To burn them, and with oxen to replace
Your gallant friends for ever. But I wish
That no man living has so long endured
The onset of his foes, as I have power
To burn or storm; for mighty Hector erst
Was slain, and now returns his safe return
Okay, that’s not Homer at his best — it doesn’t actually make sense. But it is not bad considering that it was made up by an artificial intelligence program, GPT-2, trained on the Gutenberg Poetry Corpus of pre-copyright poems by gwern (I’m a little vague on who gwern is, but I see the name on a zillion interesting/difficult projects).
As Scott pointed out earlier, AI creative writing resembles dreams.
Oddly, the one classic poet the AI program can’t imitate semi-plausibly is Alexander Pope, whose formal couplets:
NATURE and Nature’s Laws lay hid in Night:
God said, “Let Newton be!” and all was light.
were the peak of early Enlightenment poetry.
Poetry is all very fine, but I think it would be a profitable and humane undertaking to train a robot to compose corporate gobbledygook, like this AccentureOperations ad:
“SynOps ultimately showcases the art of the possible with how clients can now embrace innovation to drive new value–it’s the new applied now.”
That had to have been painful for a human to come up with, but I could imagine that an artificial intelligence entity could generate stuff like this all day long as happily as not.