Allison Goldberg introduced the enemy to her left: a humble-looking laptop running powerful generative artificial intelligence. The audience booed.
She was hosting a game show where three human comedians were competing against AI for comic supremacy. The theater was packed. Folding chairs even had to be brought in for the overflow at the Pico Theater in West Los Angeles.
“Tonight, there's gonna be some really fun moments, and then there are going to be moments that keep me up at night,” Goldberg said to the audience. “But isn't that what being human is all about?”
She used AI to place three comedians in increasingly hilarious scenarios. Goldberg, a comedian herself, came up with the concept for “Comedians vs. AI” after months of thinking about the ways technology is changing how people communicate. “AI is about to blow that up out of the water,” she says. “Perhaps mostly for bad, but maybe sometimes for good.”
With the launch of ChatGPT in November, AI has stunned people with its ability to generate information and text on a level that can seem convincingly human. But humor is a skill that even the smartest computers haven’t perfected.
Computer scientists consider humor a Turing test, or an indication of AI’s sentience, because of its complexity. Humor requires an understanding of culture, context, and linguistic nuance. Today, generative AI can create simple jokes and puns, and some robots have been programmed to perform standup comedy written by humans. But AI isn’t able to interpret sarcasm and irony the way humans can, or read a room.
At this stage in its development, AI has the language abilities of a young child, says Julia Rayz, a computer scientist at Purdue University. It can put words into logical sentences, but when it’s told a joke, it takes it literally. Rayz is studying AI to figure out why. As AI computing becomes integrated into everyday lives, she says there is a benefit to being able to communicate informally with computers and “lighten up a bit.”
While that day will arrive eventually for AI, humor is not a top research priority compared to training supercomputers to find the cure for cancer. There’s also the question of whether we want AI to be as funny as humans. Or, as Rayz says, “Will we want to just put it back in the closet, close the door, and put a big, big, big lock on it?”
The future with AI contains both opportunity and uncertainty. Headlines have mostly focused on the negative, predicting jobs lost to robots and warning of its existential threat to humanity.
“People are tired of doomsday news,” says Goldberg. Creating a show where humans compete against AI can help people process the new reality while having a good time. “There’s something very powerful in comedy,” says Goldberg.
Comedians Esteban Gast (left), Abby Feldman, and Brad Einstein represented Team Human during the comedy show. Photo courtesy of Conner Linnerooth.
Back at the comedy show, the three comedians representing Team Human were dominating the games, from improvising songs, to guessing which Florida man headline was real vs. AI-generated. ChatGPT did face one big disadvantage, though: Its programming doesn’t allow it to make fun of people.
Plus, humans like to laugh at themselves, says Abby Feldman, one of the comedians in the show.
“We have the benefit of having trauma and life experience to pull from that AI doesn't have integrated yet,” says Feldman. “And that makes us more dynamic and sensitive and hilarious for now.”
While it’s easy to imagine a time when AI can land a satisfying one liner, it’s an entirely different matter for AI to excel at live comedy, says comedian Esteban Gast, who was also in the show. Comedians build a shared language with the audience, respond in real time, and create inside jokes during a show. “At the end of the night, you're like, ‘Oh, my goodness, we went through this thing together,’” says Gast.
Goldberg’s show demonstrated that AI can, within limits, write a joke, but it’s more so just a tool to make a very entertaining live show.
That only works when you are using AI as a tool to empower people, says Brad Einstein, the third comedian in the show. He knows from experience that AI can do the opposite. Back in February, he was part of a layoff where he lost his copywriting job to AI.
“In the future, there's a huge issue about who it's going to disenfranchise,” says Einstein. “And it could very well be us, but it isn't us tonight.”