Source: Matt ReedTechnology creatives are using artificial intelligence (AI) to bring the scares this Halloween.

One such example is software designed to create masks using a neural network that was trained on 5,000 horror movie mask images. Ad agency creative Matt Reed developed the AI that teaches itself how to make the scary masks in just 24 hours.

Pitting a general adversarial network (an algorithm type in automated machine learning software) against the neural network improved the mask results over time. According to Reed, the longer the algorithm runs, the better it gets.

The mask results are unsettling, according to Reed. “What's so scary or unsettling about it is that it's not so detailed that it shows you everything. It leaves just enough open for your imagination to connect the dots."

Another way AI is helping to change the face of Halloween is through unusual costume design.

Janelle Shane, a research scientist in optics, developed a neural network capable of designing Halloween costumes. Shane employed a crowdsourced dataset for network training with results that are a “bizarre, absurd mash-up of costumes that make little sense but are hilarious to imagine."

Some of those bizarre creations include “The Spongebog,” the “Sexy Lamp” and the “Shower Witch.”

AI has also been used to liven up the soundtrack of Halloween parties everywhere this year. Creating scary songs for Halloween, Pinar Yanardag primed an algorithm with 5- to 10-second clips of horror-movie soundtracks, generating even scarier songs.

Folks wanting to end Halloween night with a scary story can do so using Botnik with its AI-generated stories, in which users can select from a list of equally scary story endings.