First of all, thanks for your patience as I disrupt my website with design experiments over the next week or so. I’m trying to see if I want to stick with WordPress hosting or migrate to another platform so pages and widgets may be jumping around or changing colors. Hopefully I won’t lose 10 years of blog posts in the process…
As I mentioned yesterday, I am devoting this week’s posts to poetry in preparation for National Poetry month in April. I am putting these posts, as well as other poetry resources, into this public Wakelet. Feel free to bookmark it or follow me on Wakelet if you are not quite ready to use them!
In today’s column, I want to investigate the idea of poetry generated using artificial intelligence – which some of us may not consider poetry. Here, as a matter of fact, is a link to an article that discusses that very question. The debate might be quite interesting to have in a secondary level class; as a hook, the teacher could display several poems and ask students to identify the ones generated by AI and the “real” ones.
The AI poetry generators out there fall into two camps: completely generated by a computer and fill-in-the-blanks. Today, I’m going to stick to the first type. Before you use any of them in class, be aware that the element of randomness means that something inappropriate may appear. For example, one of my favorite generators, Poetweet, creates poems from Twitter accounts. You don’t have to be a Twitter user to employ its craft – just type in anyone’s Twitter handle. Your students would probably want to use a famous person’s account. I tried Chrissy Teigen, who is quite prolific on Twitter, and immediately got a word I would not want to display on my projector screen at school. So, I went with my own account.
As you can see, it’s definitely not a poetic masterpiece – though fun to see what words and phrases the generator chooses.
Google has come out with a couple of poem generators in the last few years. One is, “Poem Portraits.” You “donate” a word to a crowd-sourced poem. Using machine training based on 19th century poetry, the algorithm generates a couple of lines to add to the large poem. As a bonus, you can have your “PoemPortrait” made with a picture of your contribution conforming to your face. (You don’t have to do this part – and I chose not to.) Below is my selfie-less portrait with my donated word, “splendid.”
I am more intrigued by another Google AI Experiment, Verse by Verse. There is more human interaction here, though it’s not technically a fill-in-the-blank poetry maker. You choose three poets who inspire you from a limited group of suggestions, and the type of poetry you would like to make. As you compose each verse, the AI offers suggested lines from your three poets. Here is my poem with my chosen subject, quarantine:
If you are trying to fool your students, you might want to try Poetry Ninja. I haven’t really figured out the difference between “Regular Poem” and “Mushy Poem” on this site, but when you click on the button to randomly generate (no input from you required), you might get something like this:
Many students might immediately think that this is a “real” poem due to its length and obscure vocabulary. But once you point out phrase like, “god turkey” and “banana of my disintered shoulder” they may have second thoughts…
Lastly, there is the Bored Humans Poetry Generator. This is another one that demands no work from the user other than clicking on a button – and the output reflects this.
One thing I like about the Bored Humans site is that there is a link to an article by the programmer explaining how the generator came about. There are also many, many links to other Artificial Intelligence by Bored Humans, which I have not had the chance to investigate.
Of course, I have a Wakelet so students can learn more about Artificial Intelligence. Despite its not-quite-there-yet poetry, AI is obviously becoming stronger and more prevalent, so it’s a good topic to cover the advantages and disadvantages with students.