This is the third post in a series (of two posts… so clearly I got carried away) on AI writers and their possible use in the English classroom, and education as a whole. For the previous posts, see here:
Do Androids Dream of Electric Essays? AI in the English Classroom
I, Writerbot: Critical Literacy in an Age of Machine Authors
Sidebar: shout out to Martha Wells’ series The Murderbot Diaries. Read them because they’re great. I promise I’ll stop with the robot related blog post titles now.
Story Machines
In the previous posts I explored some of the implications of AI writing in terms of students using NLP systems to write essays, and the implications of bias in AI written texts. This time I’m going to look at a more practical use case: AI storytelling and how it might be used in the classroom.
For the inspiration for this post I’m indebted to a podcast interview with Professor Mike Sharples from the UK, in which he discussed his forthcoming book ‘Story Machines: How Computers Have Become Creative Writers’. I’ve also been trawling through Sharples’ posts on AI, including this gem on the implications of AI on essay writing.
The Fox and the Lemon Tree
Fairy tales and fables are standard fare in the English classroom. The generic conventions are consistent and easy to identify, often with clear cultural markers and almost always with an identifiable moral message. I’m sure many English teachers have seen or taught a unit where students are required to read and write short fables, so I decided to use this idea to put Sharples’ Story Machine to the test.
I used the first fable-like title that popped into my head: The Fox and the Lemon Tree, and fed the Story Machine the sentence starter: “Once upon a time there lived a fox…”. Here are the three versions I generated:
| Version 1 | Version 2 | Version 3 |
| Once upon a time there lived a fox. He was very playful, always running around, jumping and climbing trees. One day, he came across a lemon tree. He looked at it for a while and then decided to climb it. He climbed up and up until he reached the top. Then he started to eat the lemons. They were so juicy and delicious! After a while, he started to feel a bit thirsty, so he climbed down and went to the river to drink some water. When he got back to the tree, he saw a big, juicy lemon waiting for him. He climbed up and ate it. It was the best lemon he had ever tasted! | Once upon a time there lived a fox who found a lemon tree. The fox was a very curious fox, and he wanted to know what a lemon tasted like. So he took a bite of the lemon, and it was the most delicious thing he had ever tasted! The fox was so excited that he wanted to share his discovery with the other animals. He went to the other animals and said, “I’ve found something wonderful! Come and try it!” But the other animals were not interested in the lemon. They thought it was too sour and they didn’t want to try it. The fox was so disappointed that he decided to keep the lemon tree to himself. He ate all the lemons and never shared them with anyone else. | Once upon a time there lived a fox who was very fond of lemons. He would often climb the lemon tree and eat the lemons to his heart’s content. One day, while the fox was eating a lemon, he thought to himself, “Wouldn’t it be great if I could have a lemon tree of my own?” So he set off in search of a lemon tree. After a long search, he finally found a lemon tree. He was so happy that he immediately started to climb it. But as he was climbing, he slipped and fell to the ground. The fox was so disappointed that he decided to go back to the lemon tree and eat the lemons. But when he got there, he found that the lemon tree was gone! The fox was so sad that he decided to never eat lemons again. |
I don’t know about you, but I like version 3. The fox goes on a seemingly unnecessary hunt for a lemon tree (maybe the first one belonged to a neighbour who resented his casual fruit larceny), and in a roundabout way the moral seems to be something along the lines of Joni Mitchell’s Big Yellow Taxi: “you don’t know what you’ve got ’til it’s gone.” The first story lacks any tension at all. I’d give it a 3/10. And the second seems to suggest that when faced with criticism the natural response is to dig in and resentfully horde lemons.
Using Story Machine in the classroom
So how would I use The Fox and the Lemon Tree – or something like it – in the classroom? Maybe something like this:
- During a unit of work on fables and fairytales, students examine a variety of texts from different cultures. They carry out close reading activities on short texts to identify generic conventions, cultural values, and morals.
- Students use the Story Machine to generate short fables, based on a title of their choice. Working in groups, they generate three to five different versions and discuss the various merits (or otherwise). They present one chosen story to the class.*
- As a class, students analyse the generated stories. Reflecting back on what they’ve learned about the cultural values in fables and fairytales, they try to culturally ‘locate’ the AI: can Story Machine’s text be placed in a particular culture? Why, or why not?
- Discussion prompt. Use the following quote from Professor Mike Sharples as a discussion point in class: “Neural network language generators [like Story Machine] are expert wordsmiths. They can produce paragraphs of plausible-looking text. But they don’t know what they write.” How is this evident in the stories you produced?
- Finally, get the students to write their own fables. The catch: they have some friendly competition in the form of Story Machine. once they have written their stories, get them to return to the AI with their title and first sentence, and see how it compares.
*For anyone worried about the inevitable use of coarse language by students, Story Machine has got your back. It has built in filters to screen for offensive language and terms. I tried to generate a Mills and Boon style romance called Hot Paris Nights (purely for research purposes) and was immediately met with this response: ** The generated text contained offensive language. Please try again with a different title or opening words. **
This is just one example of how NLP writing generators could be used in the classroom. The key takeaway here is that the technology isn’t going anywhere. GPT-3 – the technology which underpins Story Machine and many other AI writing apps – is open source, meaning anyone can access, alter, and redeploy it. Students are already using websites to “ghost write” essays, paraphrase resources, and produce writing for assessments that easily pass traditional plagiarism checkers.
These images were also generated by AI: DALL-E by Open AI
If we want students to succeed as writers in an age where they can produce texts alongside machines, we need to embrace the technology and focus on the ethics, morals, and possibilities to extend our human creativity.
I’ve preordered Professor Mike Sharples’ new book Story Machines: How Computers Have Become Creative Writers and will be interviewing the author soon. If you’re interested in learning more about the future of AI and writing, subscribe to the mailing list to stay up to date.
Leave a Reply