MIT Tried Teaching AI to Write Christmas Movie Plots

Folks at MIT have taught AI software how to create Christmas movie plots and film titles, Technology Review reports.

Inspired by Janelle Shane’s AI humor and collaboration with The New York Times to generate Halloween costume names, Karen Hao and Will Knight attempted to do something similar with AI technology and teach the system to create holiday-themed movie summaries and titles.

Using an algorithm called textgenrnn, the duo compiled over 350 Christmas movie plot summaries and fidgeted with the layers, epochs, and temperature settings in textgenrnn. Hao explains: “Layers, here, refers to the complexity of the neural network: the more layers it has, the more complicated the data it can handle. The number of epochs is the number of times it gets to look at the training data before spitting out its final results. And the temperature is like a creativity setting: the lower the temperature, the more the network will choose common words in the training data set versus those that rarely appear.”

This is what one of the summaries looked like at the beginning:

And boy son working whose issues born religious can a. Max mccallister evie to who true to her christmas the in partner question. Orphan and mccallister apartment thief most holiday.

One issue is the number of summaries Hao was using, even well over 300, they did not provide enough data for the AI to learn enough to create a meaningful summary on its own.

With a little editing by humans, here are a few plot summaries and movie title Hao and Knight were able to create using AI: