A Phrase That’s Been Stuck in My Head for Four Years, Recalled: On Ted Chiang’s Exhalation
This is not an essay about artificial intelligence; it is an essay about how great literature works.
Over the past year or so, as AI-generated art and its discontented discourse pummeled social media, an irritatingly murky phrase kept nearly announcing itself in my mind—a line from a book, no doubt—before vanishing in the corridors of my linguistically overstuffed memory. It was like an itch languishing mysteriously below the skin, unrelievable. Every time someone wondered aloud whether it was wise for humanity to develop artificial intelligence so aggressively despite what 250 years of science fiction warning us about the robot uprising, the phrase rustled into my periphery, a leaf dancing on the wind that moves it, but as soon as I reached for it, the air from my grasping hand blew it away.
Then, there it was. The two words occurred to me so randomly I at first failed to recognize its relevance. Algorithmically incompressible. Yes, those were the words; that was the phrase. But where did it come from? And why did all those ChatGPT think pieces conjure it?
Turns out it’s from a novella by Ted Chiang called “The Lifecycle of Software Objects,” from his story collection Exhalation, published in 2019. I read a galley of the book that year and fell in love with Chiang’s imaginative intelligence. He’s most well-known to mainstream audiences for “Story of Your Life,” which was the basis for Denis Villeneuve’s film Arrival, but in the sci-fi community he’s kind of a big deal. He has won four Hugo awards, four Nebula awards, and six Locus awards, all for short fiction; he has published no novels and only two story collections.
His fiction exists in the misty environs just outside our boundaries, the realm of the hypothetical. In the stories of Exhalation, he repeatedly conceives of fascinating objects that don’t exist and ponders their potential ramifications on human life, considerations Chiang meticulously unfurls with a convincing and morally complex sheen, which paradoxically renders the plausibility of his literary inventions wholly beside the point—Chiang’s gift is in making not what the device is or does seem credible (though he does) but rather the consequences of its existence on the people who use it.
Take, for instance, the story “The Truth of Fact, the Truth of Feeling,” which takes place in a future where people have been “keeping lifelogs for years, wearing personal cams that capture continuous video of their entire lives.” Although these logs represent a major cultural shift, they aren’t the agents of change in the story. Things don’t change until the release of a “new search tool” called Remem, which can sift through the recordings like a Google of your life. The algorithm “monitors your conversation for references to past events and then displays video of that event in the lower-left corner of your field of vision,” which at first seemed an endlessly useful tool (you’d never lose your phone again), but soon begins, just as the company behind the tool hoped, to “take the place of ordinary acts of recall” and become “integrated into their very thought processes.” For the narrator, who worries about the effects of such technology, the result is personally devastating: Remem proves that for many years he’s misremembered an argument he had with his daughter. He’d been convinced that his daughter had said some horrible things about her mother and his wife, who left them both years earlier, but the truth was that it was him who said those cruel things.
Chiang, though, isn’t satisfied with mere futuristic conjecture, so he juxtaposes the story of the Remem search tool with that of Jijingi, a boy in an indigenous village called Tivland, who is taught to read and write by a missionary European. Jijingi’s tribal culture is oral, so literacy and the keeping of records is foreign to them. Jijingi picks up the skills quickly, becoming the village’s scribe. In the Tiv language, Jijingi tells his teacher, there are “two words for what in your language is called ‘true.’ There is what’s right, mimi, and what’s precise, vough.” (In the parallel narrative, the narrator refers to the psychological distinction between “semantic memory—knowledge of general facts—and episodic memory, or recollection of personal experiences.”) At one point there is an inter-village dispute over ancestry, which Jijingi believes he can rightly and without conflict resolve by consulting the records of previous European missionaries. Indeed, he finds the needed information, but his village’s chief balks, asking Jijingi, “Have you studied paper so much that you’ve forgotten what it is to be Tiv?”
The narrator of the speculative parallel story characterizes things like this:
We don’t normally think of it as such, but writing is a technology, which means that a literate person is someone whose thought processes are technologically mediated. We became cognitive cyborgs as soon as we became fluent readers, and the consequences of that were profound.
Here, then, Chiang uses an already established development in history (the transition to literacy) to further his thesis on an imaginary one (a fully searchable, completely recorded life). This story is an argument, but somehow it doesn’t come across as a mere intellectual exercise. The narrator’s epiphany about the argument he had with his daughter, as well as Jijingi’s progress away from his village’s traditions and back again, elevate the story above cold, clinical analysis.
It’s clear that fictional inventions fascinate Chiang so much because he knows how fundamentally real technologies—like writing—have changed us in the past. Thus, his approach to AI in “The Lifecycle of Software Objects” is procedural, taking the reader through the development of AIs called digients, made by a company called Blue Gamma. Digients are animal-like “digital organisms” that exist only virtually in an online metaverse called Data Earth, which features “gaming” and “social continents” (italics mine), implying that it is as big as a planet. The digients can learn to talk, follow directions, and play with other virtual creatures, and they also need to eat and sleep—think of them as an incredibly elaborate Tamagotchis. Ana, a former zoo employee turned digient trainer, develops a close kinship to her pet digient Jax, “a neo-Victorian robot made of polished copper.” When another company creates real humanoid robots with display screens for heads that the digients can wear—and thus seem like actual creatures moving through the world—things begin to get morally complex. Are these digients alive? What rights do they have? The people who love them certainly treat them like living beings, and numerous corporations make big profits from their life-like qualities, so perhaps they merit some legal protections.
As Ana tries to navigate the changing technologies that support Jax, she finds herself at a crossroads with the wealthy businesses that must fund the genomic engines the digients exist within. The companies eventually want to use the technology to create an army of docile employees, but who aren’t “owed the same obligations” as people. But Ana knows that this is an impossibility, and she wants them to understand that
…there are no shortcuts; if you want to create the common sense that comes from twenty years of being in the world, you need to devote twenty years to the task. You can’t assemble an equivalent collection of heuristics in less time; experience is algorithmically incompressible.
And there it is, the phrase, completed now: “experience is algorithmically incompressible.” Yes, that’s it.
These words have clanked around my head for the past four years, taken from a piece of fiction and blurrily remaining, ready to be evoked at the appropriate moment. Somehow, though I nearly forgot its source and definitely lost its context, this phrase not only stuck around, it also arrived unbidden at the exact correct stimuli. Surfacing only when I was confronted by the recent AI discourse, which is remarkable because I wasn’t even conscious that the phrase was related to the subject. I mean, obviously I subconsciously knew its meaning, but not articulately.
But literature gave me this idea—it is not an insight I am capable of making—and now I can never unknow it. Moreover, the idea that human experience cannot be replicated in digital form provides me with some comfort when thinking about the bewildering changes new innovations are likely to bring about. I don’t know anything about AI or its potentiality, so I could be completely wrong to feel assured by such a notion, but even if that were true, I wouldn’t blame literature—or Chiang specifically—for misleading me. No, Chiang’s insight didn’t stick with me because I wanted it to; it stuck with me because exceptionally well-put phrases always stick with me. I had never thought of experience in quite that way before, and I definitely had never considered its application to the development of technology. But I think the thing that really got the phrase caught in my head is just how wonderfully concise it is. Try thinking of two other words that communicate the complexities of this idea better than “algorithmically incompressible.” If you can, you’re much more adept at language than I am.
Another beautiful function of literature: this phrase and its connotations remind me of another line from another great work of science fiction. At the end of H.G. Wells’s The War of the Worlds, the invading Martians perish because their physiology hasn’t been prepared to deal with “putrefactive and disease bacteria” that live on Earth. As the narrator explains:
But there are no bacteria in Mars, and directly these invaders arrived, directly they drank and fed, our microscopic allies began to work their overthrow. Already when I watched them they were irrevocably doomed, dying and rotting even as they went to and fro. It was inevitable. By the toll of a billion deaths man has bought his birthright of the earth, and it is his against all comers; it would still be his were the Martians ten times as mighty as they are. For neither do men live nor die in vain.
As germs protected humanity from an alien invasion, so too will experience guard us against the robot uprising. Experience is like a weapon I didn’t realize humanity possessed, just as Chiang had provided me with an intellectual response to an emerging technological development, almost entirely unbeknownst to me. I thought I was simply liking a phrase, but what I was really doing was ingesting a belief about the impossibility of bottling wisdom, which is ironic because that is precisely what Chiang had done. Perhaps, then, it’s not containing wisdom that’s impossible but passing it on, the way a song is easy to hear but hard to play by ear. Because even though I loved Chiang’s succinct and unique phrasing, noted it, and was able to recall it all this time later, I didn’t learn the depths of its meaning, or the reach of its philosophical implications, or the nature of its comfort, until now, four years after encountering it. Like the way the Tiv language in that other Chiang story has two words for ‘true,’ one referring to “what’s right” and another to “what’s precise,” literature can impart its truths to you via precision, elegant and witty and edifying eloquence so pleasing it’s effortlessly remembered, but the actual wisdom contained in the eloquence isn’t revealed until it’s relevant to your life, until you see the phrase not for its form but its function, that it isn’t writing as much as it’s communication. You must learn for yourself because experience is algorithmically incom-fucking-pressible. I’m sorry I keep repeating it, but there truly is no better way to say it, and that—that is how great literature works.
Jonathan Russell Clark is a former staff writer for Literary Hub, and his essays and reviews have appeared in the New York Times Book Review, the San Francisco Chronicle, The Atlantic, Tin House, Vulture, the Columbus Dispatch, LA Review of Books, New Republic, The Georgia Review, and dozens of others. Most recently, he wrote about a debut novel by a professional skateboarder for the L.A. Times. His first book, An Oasis of Horror in a Desert of Boredom, a study of Roberto Bolaño’s 2666, was published in 2018. His second book, Skateboard, will be a part of Bloomsbury’s Object Lessons series and will be published in 2022.