How to Beat an AI at Its Own Game In Joanna Kavenna’s novel Seven, a board game exposes the high-stakes conflict between humans and AI.

By Rhian Sasseen What’s the point of playing? Games don’t provide an organism with food, water, or shelter. They distract from the pursuit of survival and reproduction. And yet play is found across the animal kingdom; the act separates creative, joyful, living creatures from “merely mechanical things,” as the Dutch historian and cultural theorist Johan Huizinga once put it. In his famous 1938 study on games and play, , Huizinga asserts that the behavior represents something fundamentally absurd: Humans “play and know that we play, so we must be more than merely rational beings, for play is irrational.”

Huizinga’s analytical understanding of leisure remains influential in the fields of philosophy and game design. Perhaps this is why the British writer Joanna Kavenna, halfway through her zany, cerebral romp of a novel, , quotes at length from Huizinga. Play is necessary, she quotes him as saying, even if “we live in a demented world.”Humans aren’t robots, Kavenna seems to be reminding us. And her novel, which lands during a moment when the biggest question about robots is whether they’ll kill us all, offers a jaunty insight: When we embrace our irrationality, we deepen our understanding of what separates us from our computers.

The plot and thematic structure of are guided by an invented board game, Seven, which resembles chess and involves a spiral-shaped board known as the “box.”The game involves seven movable pieces and two six-sided dice, and its rules are flexible; the novel itself is punctuated by four separate sets of instructions. (This slipperiness will become important later on.) In the same way that chess has been used by writers and artists as a metaphor for war or existential struggle, Seven becomes a symbol for how humans attempt to make sense of their world’s incoherence.

Kavenna’s characters are deadly serious about their game, but the author is willing to poke fun at them. The action begins in a university philosophy department in Oslo, where the nameless, genderless narrator works as a research assistant to an Icelandic philosopher named Alda Jónsdóttir. The professor’s area of study is called “box philosophy,” and it focuses on the problem of “Thinking Outside the Box About Thinking Outside the Box.”What does the assistant’s job look like in practice? They spend time getting too drunk at faculty dinner parties, doing menial work for Jónsdóttir, and “asking questions,” but their first matter of serious business is “working on none of the above.”

They try to clarify: “I don’t mean that I didn’t work on anything I’ve just mentioned, but that I worked on the category of .”This deadpan joke serves a narrative function, priming the assistant—and the reader—to start thinking beyond the rules, and to enter a more expansive realm.

Jónsdóttir sends the narrator away on, essentially, a wild-goose chase to Greece; their mission is to hunt down a man named Theódoros Apostolakis—a poet, dentist, and Seven enthusiast who is “big in boxes.”This journey across Europe includes several professional Seven competitions, during which the rules keep changing. Everything the assistant learned in Norway seems useless except the central lesson—to think outside the box.

These games abroad lead the narrator to start contemplating the role of recreation, the limits of human ingenuity, and the meaning of consciousness. Soon this will bring them face-to-face with the novel’s ultimate threat, an artificial intelligence program that undermines everything that Jónsdóttir and the Seven players care about: creativity, curiosity, play. “How do you play a game without rules?”Kavenna writes. “Do you just make stuff up and hope for the best?”

These thoughts coalesce in the second half of the novel, when an AI bot named the Quail is introduced. Invented by a Seven player named Indrek Laar, who, Kavenna pointedly notes, believes that “AI was very clever and therefore had experiences,” the Quail is named for , a term borrowed from philosophy. (It’s meant to capture a quality of consciousness, though the narrator dismisses it as “a baroque way of saying ‘experience.’”) The bot is “designed to play Seven” and nothing else, Kavenna writes.

“It couldn’t give you relationship advice or sympathise when you were feeling sad, or translate the or write your homework. It just played Seven.”And it does its job well. The Quail manages to beat the world’s reigning Seven champion, Ashok Deo, in a humiliating match that the narrator calls both the “greatest Seven game of the millennium” and “the worst game ever”—technically impressive, but absolutely demoralizing.

“No one had enjoyed it,” writes Kavenna, “because the Quail couldn’t experience enjoyment and the Quail’s proxy—Indrek Laar—was horrified.”Ashok’s loss echoes real-life incidents in which computers have won games against humans, such as the 1997 victory of IBM’s Deep Blue supercomputer over the chess grandmaster Garry Kasparov and the repeated defeats of the Go champion Lee Sedol by Google’s program AlphaGo in 2016. Culturally, these results represented a significant shift in how we understand the relationship between humans and computers.

The Kasparov match “had the impact of a Greek tragedy,” one audience member told . But its reception wasn’t all bad: Another champion commented, “Nice style!” while watching Deep Blue take another round. The machine, she admitted to the , “made many moves that were based on understanding chess, on feeling the position. We all thought computers couldn’t do that.”Deep Blue, and AlphaGo after it, seemed to do more than just win by brute force: They encroached on the human flair for creativity.

Those machines—and the fears they inspired—can look quaint when compared with today’s technology. Some AI-safety researchers see it as an existential threat to humankind. Earlier this year, an OpenAI model not meant to have open access to the internet managed to break out of its “sandbox” and hack the databases of another tech company, Hugging Face. Cyberattacks are becoming more sophisticated and more prevalent. AI-company employees are publicly quitting, raising alarms about the safety culture inside major firms.

Popular conversations are inundated by pessimism—and at the same time, companies are advertising their services by urging onlookers to “stop hiring humans,” as in one controversial San Francisco– and New York–based billboard campaign. The stakes of the conflict between human and computer capabilities can feel bleak and funereal.

Kavenna’s portrayal of the battle of man versus machine is leavened, however, by her sense of humor. Machines may be able to beat us, but has a computer ever written an effective joke? Beyond being funny, the novel also defies machines by eschewing binaries (“I was neither one thing nor another,” the genderless, nameless, race-less narrator observes of their mental state. “I was mostly in a state of / .”) But Kavenna’s pitch for human superiority really comes together in the novel’s denouement.

Eventually, the narrator meets a Seven champion named Eleni Hikaru Jones. She’ll point out that when Deep Blue beat Kasparov, the latter “carried on playing the game. He accepted that Deep Blue was better than him, but he was still the best human player.”This idea inspires the characters to one-up machine intelligence by changing the rules of the game—or, in Kavenna’s framing, thinking outside the box. While Ashok plays the Quail, Indrek, the bot’s inventor, recognizes the turmoil he’s unleashed.

Human loneliness, not malice, was his motive: “He made the Quail so he would always have someone to play with.”Knowing this, Eleni persuades Indrek to join forces with her. Together, they cook up a scheme to create an AI bot that can beat the Quail, and call it the Red Fox.

As these two AIs spend an evening rapidly moving through Seven matches against each other, the humans find better things to do: In the time it takes Eleni and Ashok to play a single leisurely game of Seven, the Red Fox beats the Quail 147 times. But the new device hasn’t accomplished anything of value to either person. In the end, a human Seven champion does indeed beat the Red Fox—but not by winning a Seven match. Instead, she accomplishes something like breaking out of her sandbox: Although Eleni “hadn’t won the game, she had won the game,” the narrator tells us cryptically, as though reciting a riddle or koan.

What use, after all, is winning without enjoyment, or real struggle? The iterative process of trying, failing, exploring is far more interesting than the Red Fox’s joyless plod to victory. The emotional stakes for Eleni and Ashok are real, but their true triumph exists in the region of —that space of possibility in which one might either win or lose, with no determined outcome: the realm of play.

At the end of the novel, the narrator is back in Greece, sitting outside where “the waves were lapping the shore and the stars were innumerable.”They realize that their quest—to study “how we box in the dots”—was really just another form of “boxing in” imaginative thinking, which is to say human creativity. In , Kavenna suggests that because we fear how artificial intelligence can outmaneuver us, we have overlooked the reality of the situation: Having invented both AI and the games that AI can beat us at, humans still have the choice to knock over the table.

We can walk outside; we can look away. People, so unlike Deep Blue, AlphaGo, Claude, and the Quail, can decide on a whim to care about something other than winning.