ChatGPT turns children's book into infinite loop social commentary

Our second son, Cal, has a beloved stuffed monkey. Monkey goes with Cal to bed, to breakfast, on the ride to school, and when we go on vacation. So it was no surprise this morning when I heard my wife and children commenting that they wished the beloved children's author Laura Numeroff had created a version of "If You Give a Mouse a Cookie" but with a monkey and a banana.

(For those unfamiliar, the story line is that if you give a mouse a cookie, he'll want milk to go with it, which then leads to him wanting other things, which eventually leads back to him wanting a cookie all over again.)

Like any good father in 2024, I immediately fired up my chatGPT terminal and fired in the prompt to create such a story. The result both surprised and amused me.

At first, it wrote a cute story about a monkey learning to share with others. But it wasn't the same circular story line as IYGAMAC. So I corrected it with a mildly derogatory question and had it try again. It started typing. I waited and skimmed the story before realizing it was repeating itself. It had failed to learn that the repetitive nature of the story is implied, not explicit. So around and around it went, stuck in an infinite story-telling loop. This is interesting in and of itself; folks have achieved infinite loops in chatGPT using Fibonacci series and other complex mathematical approaches, but this was entirely literary.

The other fascinating part to me is the mechanism it chose to lock our mythical monkey into an infinite loop: social media. Just as a mouse cannot refuse a cookie, so AI has determined social media to be irresistible. Our dear monkey makes it only to paragraph three before he's posting his first selfie. From there, it's game over. Notifications, likes, profiles, messages, followers, profiles, likes and around and around it goes. He never escapes.

He never eats another banana. In this story, the monkey starves to death over time, never balancing his social media consumption with his physical consumption.

In the vastness of documented human knowledge, AI sought out a repetitive and addictive task around which it could build a circular story line and the winner (loser?) was...social media.

My full conversation with chatGPT available here.

What does this tell us about AI's perception of social media?

What actions could AI take against social media based on this perception?

If humans lack the self-control necessary for our utmost happiness, should we rely on AI to take action on our behalf? The plot of iRobot comes to mind.

Let me know your thoughts in the comments. I'm finding AI continues to surprise and delight me as we learn more about this thing we created yet cannot fully explain.