Home Tabletop Cards against humanity put its writers against AI to see who comes up with the best cards

Cards against humanity put its writers against AI to see who comes up with the best cards

3 min read
10

There are many industries and professions that are trying to diversify and figure out how to survive with the rise of artificial intelligence and machines that could take their jobs. And while several occupations that rely on repetitive tasks are susceptible to being made obsolete by our future robot overlords, it is commonly believed that the one avenue where humans are most likely to maintain their jobs is in the realm of creativity. Or at least, that’s what we would like to believe.

The creators of Cards Against Humanity decided to put this notion to the test during a Black Friday stunt where they created a machine-learning algorithm that can write cards for the game and then put their team of writers to the test to see who could create the better cards. The company allowed people to go onto its website where they could choose to upvote or downvote certain cards on each side before choosing to buy, while also live-streaming the event where you can watch the writers struggle to come up with fresh ideas.

On the line are $5,000 bonuses for every employee if team human comes up victorious, or heartless termination in the event the AI takes the top spot. I don’t think the company actually plans to fire their writers if they lose, but it is a clever stunt nonetheless to drum up the human versus machine narrative while making a point for the gross commercialisation of the day. As revealed in a statement last year by co-creator Max Temkin when they also proposed a similar bizarre activity for the day:

Black Friday probably represents the worst things about our culture. It’s this really repulsive consumerist frenzy right after a day about being thankful for what you have. So it’s always seemed like a really good subject for parody to us.

To create this event, the company made use of a neural network from the open source GPT-2 model created by OpenAI which is already trained on roughly 40,000 books worth of internet text to ensure it can reliably predict and fill out the next word or punctuation mark in a sentence with realistic effect. They then fed it an additional 44,000 cards from the game (2,000 cards the feature in the official game, another 25,000 internal brainstorming cards that never made it into the game, and 17,000 unofficial cards from fan-curated lists). That is a massive advantage that should give the AI a massive advantage on which to draw upon.

In the end though despite the cleverness of the AI, the humans came out on top (albeit narrowly) making $82,860 versus the AI’s $81,135, proving not only that AI is not ready yet to take their jobs – though perhaps not far away – but also ensuring they pocket an extra $5000.

Last Updated: December 2, 2019

Check Also

AI Dungeon 2 has been released, showcasing even more random, machine-powered adventures

Ever thought AI would make a decent dungeon master? Well, AI Dungeon 2 is continuing to wo…