He Studied Cognitive Science at Stanford. Then He Wrote a Startling Play About A.I. Authoritarianism.
When I saw “Data,” a zippy Off Broadway play about the ethical crises of employees at a Palantir-like A.I. company, last month, I was struck by its prescience. It’s about a brilliant, conflicted computer programmer pulled into a secret project — stop reading here if you want to avoid spoilers — to win a Department of Homeland Security contract for a database tracking immigrants. A brisk theatrical thriller, the play perfectly captures the slick, grandiose language with which tech titans justify their potentially totalitarian projects to the public and perhaps to themselves.
“Data is the language of our time,” says a data analytics manager named Alex, sounding a lot like the Palantir chief Alex Karp. “And like all languages, its narratives will be written by the victors. So if those fluent in the language don’t help democracy flourish, we hurt it. And if we don’t win this contract, someone else less fluent will.”
I’m always on the lookout for art that tries to make sense of our careening, crises-ridden political moment, and found the play invigorating. But over the last two weeks, as events in the real world have come to echo some of the plot points in “Data,” it’s started to seem almost prophetic.
Its protagonist, Maneesh, has created an algorithm with frighteningly accurate predictive powers. When I saw the play, I had no idea whether such technology was really on the horizon. But this week, The Atlantic reported on Mantic, a start-up whose A.I. engine outperforms many of the best human forecasters across domains from politics to sports to entertainment.
I also wondered how many of the people unleashing A.I. tools on us really share the angst of Maneesh and his co-worker, Riley, who laments, “I come here every day and I make the world a worse place.” That’s what I think most people who work on A.I. are doing, but it was hard to imagine that many of them think that, immersed as they are in a culture that lauds them as heroic explorers on the cusp of awe-inspiring breakthroughs in human — or maybe post-human — possibility. As a New York magazine review of “Data” put it, “Who gets so far at work without thinking through — and long since justifying — the consequences?”
Sign up for the Opinion Today newsletter Get expert analysis of the news and a guide to the big ideas shaping the world every weekday morning.
But last week, Mrinank Sharma, a safety researcher at Anthropic, quit with the sort of open letter that would have seemed wildly overwrought in a theatrical script. “The world is in peril,” he wrote, describing constant pressure at work “to set aside what matters most.” Henceforth, said Sharma, he would devote himself to “community building” and poetry. Two days later Zoë Hitzig, a researcher at OpenAI, announced her resignation in The New York Times, describing the way the tool could use people’s intimate data to target them with ads.
I reached out to the writer of “Data,” Matthew Libby, because I was curious about how he got so much so right, and learned that before he worked in theater, he studied cognitive science at Stanford. More specifically, he has a degree in symbolic systems, an interdisciplinary program that combines subjects including computer science, philosophy and psychology. He always intended to be a writer, he said, but wanted to make sure he had something to write about.
Not surprisingly, Libby, who graduated in 2017, felt the pull of Silicon Valley, at one point interviewing for an internship at Palantir. He was heartbroken when he didn’t get it. But when he came across a 2017 Intercept story headlined “Palantir Provides the Engine for Donald Trump’s Deportation Machine,” he wondered what he would have done if he’d worked there, which is how “Data” was born.
Perhaps the most interesting thing about “Data” isn’t its insight into those who leave companies making dangerous A.I., but into the majority who stay, and the stories they tell themselves about what they’re building. “My experience of the tech industry is just that there’s always this air of inevitability,” said Libby. “You know, ‘We can’t pause any of this because it’s coming no matter what, and don’t you want to be the person doing it?’”
Among technologies, A.I. is unique in that those who are creating it — and profiting off it — will from time to time warn that it could destroy humanity. As Sam Altman said in 2015, shortly before helping found OpenAI, “I think that A.I. will probably, most likely, sort of lead to the end of the world. But in the meantime, there will be great companies created with serious machine learning.” A slightly truncated version of that quote appears as an epigraph in Libby’s script.
Just last month Dario Amodei, who leads Anthropic, the most seemingly responsible of the A.I. giants, published an essay titled “The Adolescence of Technology,” about potential A.I. apocalypses. A.I. systems, he wrote, could turn against humankind or help to create biological weapons. They could be used to build a digital panopticon more comprehensive than anything existing today, or develop propaganda so precisely tailored to its users that it would amount to brainwashing.
But as Amodei sees it, these hellish possibilities are less reasons to slow A.I. development, or to keep it out of the hands of the surveillance state, than to make sure that the United States stays ahead of China. “It makes sense to use A.I. to empower democracies to resist autocracies,” he wrote. “This is the reason Anthropic considers it important to provide A.I. to the intelligence and defense communities in the U.S. and its democratic allies.” His argument would be sounder if the United States was still, in any meaningful sense, part of a coalition of democracies, rather than a nation ruled by an aspiring autocrat who is propped up in no small part by the tech industry.
In “Data,” Alex makes a similar argument for bidding on the Department of Homeland Security contract. “We’re the fighters protecting democracy,” he says. “China already has an automated social credit system they’re exporting to developing nations. Russia has the most targeted disinformation infrastructure known to man. That’s what they’re innovating towards. If we stop innovating? We lose our lead.” The threat of authoritarianism abroad becomes a rationale for building the tools of digital authoritarianism at home. Too bad it’s not just fiction.
Michelle Goldberg has been an Opinion columnist since 2017. She is the author of several books about politics, religion and women’s rights and was part of a team that won a Pulitzer Prize for public service in 2018 for reporting on workplace sexual harassment.