The creators of Cards Against Humanity are back for their annual Black Friday stunt, and this one is delightfully dystopian. Starting at 11AM ET today and lasting for the next 16 hours, the human writers on the CAH team are facing off against an artificial intelligence to see who can create the most popular new pack of cards, based on how many people pay for more $5 packs. You can upvote or downvote your favorite cards for each side on CAH’s website before buying, and you can also watch the humans struggle to come up with new iterations in real time over live stream.
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. We don’t think CAH actually plans to fire their writers if they lose, but it is a clever stunt nonetheless to drum up the human vs. machine narrative at a time when automation may pose a very real threat to millions of jobs in the coming decade, writing included.
It follows the company’s tradition of pulling Nathan For You-style capitalism parodies on the most commercial day of the American calendar year. Last year, CAH held a 99 percent off sale on a series of outlandish items like a 17th-century halberd and a 2015 Ford Fiesta with just 25,000 miles on it. (The company reportedly did ship some of the items in the sale, at least those that were sourced from its own office.) In 2013, the company raised the price of its card packs by 100 percent, just because it could.
“Black Friday probably represents the worst things about our culture,” Cards Against Humanity co-creator Max Temkin said in a statement last year. “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.”
Cards Against Humanity built a real neural network using a model created by OpenAI
And in keeping with CAH’s absurd commitment to the gag, the AI isn’t just some random text generator. It’s a legitimate neural network, borrowed from the open source GPT-2 model created by AI research company OpenAI and trained specifically to write CAH cards. The GPT-2 model 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. But then CAH went further and trained it using tens of thousands of its own cards.
“This was done by taking that pre-trained network and then training it further on the text of 44,000 white cards. That includes all (roughly) 2,000 cards 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,” reads the in-depth explanation of the AI on CAH’s website.
“We stopped the training once it could “consistently” produce cards matching the grammar and tone of the game,” it goes on. “We did this so it didn’t draw too much from already-written cards while still leveraging as much cultural information as possible.” After that, CAH ran a straightforward filtering algorithm to ensure it could pluck out AI-generated cards that matched its standard format and to avoid ones that were too similar to existing cards.
As it stands right now, the human team is narrowly beating out the AI by just $700 or so, with some gems like “Sucking all the oil out of the planet and fucking off to Mars” and “That whole Jeffrey Epstein thing.”
Vox Media has affiliate partnerships. These do not influence editorial content, though Vox Media may earn commissions for products purchased via affiliate links. For more information, see our ethics policy.