Silicon Valley? More Like…Liberal Arts Valley: Meet the English Majors Behind Anthropic

When OpenAI released ChatGPT to the general public on November 30, 2022, it touched off the generative AI boom and established itself as the undisputed frontrunner in the race toward artificial intelligence hegemony. With unprecedented speed, the chatbot swept across the world, penetrating its many markets, answering prompts in places as remote as Svalbard and Nauru. ChatGPT became, at that comparatively innocent and quiet moment in time, the fastest-growing consumer application in history. And as a result, OpenAI’s worth soared, and soon its position as the most valuable AI startup seemed unassailably secure.

This was a remarkable feat given that OpenAI had only been launched seven years earlier, after Elon Musk and Sam Altman, concerned with the unchallenged growth of Google’s artificial intelligence division, had linked up to create the non-profit whose express purpose was to make sure that AI ultimately benefited humanity. (OpenAI has since undergone a kind of corporate surgery and has, for all intents and purposes, become a for-profit enterprise.)

For a while OpenAI commanded a huge lead over the other big players in the field of artificial intelligence. Like in so many other industries, the race toward AI supremacy seemed to be a sprint, and sprints hinge on speed, and OpenAI had demonstrated its swiftness by releasing its consumer chatbot first and beating its rivals, especially Google, to the punch. But when OpenAI was unseated last month as the hottest name in artificial intelligence, it wasn’t by Google, the primordial threat. And it wasn’t by Meta, whose CEO, Mark Zuckerberg, has plowed enormous sums of money into his struggling AI division. And it certainly wasn’t unseated by Elon Musk, who has had more success, through his Colossus One and Colossus Two data centers, at renting out compute than in his own artificial intelligence ventures.

No, it was Anthropic that summarily dethroned OpenAI and became the hottest name in artificial intelligence after it raised $65 billion in financing from investors who valued the startup at $965 billion. To put that number in perspective, ByteDance, the Chinese firm behind the unnervingly omnipresent TikTok, has an estimated valuation of $550 billion. And to put that number in further perspective, and in a way that makes sense to a lout like me, Buckingham Palace is believed to be valued at around $5 billion. That makes Anthropic worth approximately 193 Buckingham Palaces. And OpenAI, according to its latest fundraising round in March, is worth $852 billion which, using my preferred metric (the Goodbar metric, if you will), makes it equivalent to about 170.4 Buckingham Palaces. (This raises the crucial question of what, exactly, is .4 of a Buckingham Palace, which I will address in a future post.)

Of course, the artificial intelligence race is far from over. These valuations—decidedly unmoored from anything as mundane as revenue or profit—are liable to rise or fall depending on the market’s mood. It is unclear which AI firm will ultimately emerge as victorious. But, given Anthropic’s current ascendancy, I think it’s worth taking a look at the company that’s managed to reach the top of the world’s most competitive industry in just five years.

The San Francisco startup was born in 2021, when a group of seven OpenAI employees, disenchanted with that organization’s increasingly ambivalent attitude toward AI safety, left and formed their own company. As one might expect, some of the founders possess the traditional San Francisco-startup pedigree. Earlier in his career, Anthropic CEO Dario Amodei cut his teeth as a deep-learning researcher at Google Brain, where another founder, Chris Olah, also worked. Tom Brown, the chief compute officer, supplies the requisite MIT and Y Combinator backgrounds.

But what’s striking is that, among these seven OpenAI defectors, there aren’t only AI researchers and highly educated scientists. In fact, two of the founders, Jack Clark and Daniela Amodei (Dario’s sister), share the unique distinction of being tech leaders who studied literature in college.

Clark, whose official title is Head of Policy, attended the University of East Anglia, in England, where he majored in English literature with Creative Writing. After graduating, Clark worked a series of journalism jobs, eventually distinguishing himself as a tech reporter. In 2016, he pivoted and took a position at OpenAI, where he ultimately became a policy director.

Clark is not ashamed of his non-tech background and openly considers his liberal arts education of great value. Indeed, in some ways, he almost seems to view his decision to study literature as being acutely prescient. In April, at the Semafor World Economy Summit, he said:

I’m a literature graduate, and I don’t think you’d put that as a co-founder of a frontier AI company. But what turned out to be useful is I got to learn a lot about the stories that we tell ourselves about the future. That’s turned out to be extremely relevant for AI in a way that I think people wouldn’t have predicted.

The Anthropic co-founder then elaborated on which majors, in a post-artificial intelligence world, would be worth pursuing. And while he didn’t name any specific areas of study, his description highlighted skills and abilities traditionally associated with liberal arts majors: “Majors that are going to become more important are ones which involve synthesis across a whole variety of subjects and analytical thinking.”

Daniela Amodei, the president of Anthropic, studied English literature at UC Santa Cruz. Like Clark, she seems to view her liberal arts background as singularly indispensable in the post-AI world. And like Clark, she considers “critical thinking” to be a skill of paramount importance. In an ABC interview, after she was asked what she majored in, Amodei stoutly championed the humanities:

I ultimately believe in a world where AI is very smart and capable of doing so many things, the things that make us human will become much more important instead of much less important…. I actually think studying the humanities is going to be more important than ever. A lot of these models are actually very good at STEM.

Amodei indicated that such considerations inform Anthropic’s hiring practices, which prioritize candidates who “are great communicators, who have excellent EQ and people skills, who are kind and compassionate.” Anthropic’s president concluded with an observation that, while more technical skills might be imperiled by AI, the ability to understand humanity at a fundamental level—and that is an ability—will remain invaluable:

“There are things that make us uniquely human, understanding ourselves, understanding history, um, understanding what makes us tick, I think that will always be really, really important.”

I’d argue that, at Anthropic, this culture of humanities doesn’t begin and end with Daniela Amodei and Jack Clark. It seems to have suffused other areas of the company. Its roughly 20,000-word constitution, which outlines Claude’s (its chatbot’s) values—thereby giving it moral ballast—strikes me as a fitting move for a startup concerned with the question of what makes us “uniquely human.” Anthropic also bolstered its humanities credentials when it hired Amanda Askell, a Scottish-born and Oxford-educated philosopher, and made her head of its personality alignment team. In this role, Askell has helped build what might poetically be called Claude’s soul or, if you are of a more prosaic slant, the complex ethical framework with which the chatbot answers users’ questions.

For further evidence of a culture of humanities, look at CEO Dario Amodei, who might have a scientific background but whose considerable literary output is something to be acknowledged and even admired. Across two long-form essays, together weighing in at around 34,000 words (he’s a man after my own heart), Amodei has outlined his vision, in exquisite detail, of artificial intelligence’s potential risks and rewards. Then there’s co-founder Chris Olah, who notably endorsed Pope Leo’s encyclical warning of AI’s implicit and explicit dangers to humanity. I’d argue that this, too, was a humanities-minded maneuver.

The broader battle between the different AI companies is far from over. And the showdown between OpenAI and Anthropic, which is a somewhat internecine conflict given Anthropic’s origin, is only just heating up. There is no love lost between the San Francisco rivals, and the feud could devolve into one as bitter and public as that between Jay-Z and Nas. (That sentence was for some additional street credit or “cred” as some of us are wont to say in more informal settings.) OpenAI and Anthropic espouse corporate philosophies that are, in essence, at odds with each other. While Anthropic has positioned itself as the more circumspect firm, OpenAI has become the necessary foil, one that stresses the need to maintain the competitive edge over China, which means that, in certain instances, safety considerations are of lesser consequence.

Given that poll after poll reveals that an increasing number of people are anxious about artificial intelligence, I think that the AI company that is the most humane, in both senses of the word, will be the one to carry the day. For the moment, the most humane AI startup appears to be Anthropic. It is the most valuable one, too.