At times, it’s possible for an audience member at Matthew Gasda’s play, Doomers, to feel like a fly on the wall in a San Francisco startup “war room.” Characters bicker about product releases and name-drop venture firms, only breaking to have boba or tacos delivered. Then a polycule joke or a weighty monologue will return one to the satire at hand in the small Manhattan art gallery where the play is being staged.
Doomers is maybe the first artistic dramatization of events that rocked the tech industry in the fall of 2023, when OpenAI’s board briefly fired CEO Sam Altman over AI safety and interpersonal concerns. The play is very loosely based on the hours after Altman was fired, though all of the names have been slightly changed and some characters are composites or archetypes.
Billed as “a Glengarry Glen Ross for the AI age,” the action takes place in two closed-room acts—one featuring the company’s inner circle and the other, the board. Through Succession-esque banter, they each reflect on humanity’s AI future and increasingly loony doomsday scenarios. That philosophizing is set amid interpersonal squabbling, status jockeying, and plenty of satire of the San Francisco AI scene’s cultish feel.
The real-life drama could be read as the last stand of OpenAI’s nonprofit ethos against a tide of money or a rogue board on a misguided quest. Gasda doesn’t lean too much into either of these interpretations. Most of the characters involved are ambitious and self-interested, often bumbling or cynical, with occasional flashes of real emotion or ideals.
“I basically really believe that my job as a playwright is to show the psychological realities, to show—and it’s a cliché—the human side of these people who, in a sense, don’t even fully believe themselves to have a human side,” Gasda told Tech Brew. “I did want to show that they get hungry, they fart, they have needs, that they’re insecure.”
The production begins with five actors playing members of the higher-ups at MindMesh—the not-so-subtle OpenAI analogue—hunkered down after they’ve received the news of the board firing its leader, Seth. Two camps emerge in a debate over how to proceed with Seth and Myra—a stand-in for former OpenAI CTO Mira Murati—on one side and the company’s chief ethics officer and a Stanford AI lawyer on the other.
Keep up with the innovative tech transforming business
Tech Brew keeps business leaders up-to-date on the latest innovations, automation advances, policy shifts, and more, so they can make informed decisions about tech.
“I was fired for creating miracles,” Seth says at one point.
The next act shows the Celsius-fueled board members, with some even clearer creative liberties taken. Stand-ins for Quora CEO Adam D’Angelo and board member Helen Toner are joined by a portrayal of AI writer Eliezer Yudkowsky (not an actual board member), a jaded venture capitalist who thinks AI is only good for porn, and a 25-year-old founder taking magic mushrooms.
“A lot of the board’s real-life people were quite opaque,” Gasda said. “Because it was a comedy, I wanted to expand the parameters of who was there.”
Gasda said he already had the seed of an idea for the play when he had an hours-long conversation at a party with some AI insiders about possible “p-doom scenarios”—the AI industry’s half-joking term for the probability of various apocalyptic outcomes of AI. Some that made it into the play involved “the probability that aliens are guarding the planet, and stepping in to stop the nukes, and the probability that AI will commit suicide if it gets too smart.”
“There are these people at a party who are otherwise, totally just like everybody else, but they’re planning for the end of the world, and maybe they themselves are working on something that they think might destroy the world and destroy civilization,” Gasda said. “And I was like, ‘That’s a crazy thing to wake up and go to work and do every day.’”
The play program credits Anthropic’s Claude and OpenAI’s ChatGPT with providing dramaturg, and Gasda said he did rely on AI for asking background questions and generating names. There’s also a scene toward the end of the play where the MindMesh chatbot is prompted to provide an allegory of itself. The response it gives, comparing itself to the “gardener of Babylon,” was actually generated by Anthropic’s Claude and makes for a big chunk of a key monologue.
As for whether any of the real-life people involved have seen the play, Gasda can’t say for sure, but he has heard that someone has sent Altman himself a copy of the script.
“I wasn’t fully expecting to send it to Sam Altman,” Gasda said. “I don’t know if he’s going to read it or have an assistant—probably an assistant got it.”