
In the late ‘80s, Wendell Berry declared jihad against the computer age with a short but controversial essay in Harper’s Magazine. “Why I Am Not Going to Buy a Computer” seemed like the odd ravings of a crank when I read it as a college senior, but also deeply compelling for reasons I would only be able to articulate later in life. To put it bluntly, I was not a ready-made audience for this essay. My father was a computer engineer for Bell Laboratories, and we were the first house in our neighborhood to own a personal computer in the late 1970s. My uncle was a mechanical engineer who worked on the Hubble Space Telescope. We were a technology-forward family living in the halo of 1960s techno-utopianism. Computers were the future. Why would anyone not want to buy a computer in 1987?
Berry, a long-time Kentucky farmer, poet, essayist, and environmental activist, opens his essay with this declaration of purpose: “Like everybody else, I am hooked to the energy corporations. I hope to become less hooked to them. In my work, I try to be as little hooked to them as possible. As a farmer, I do all of my work with horses. As a writer, I work with a pencil or pen and a piece of paper.” If Henry David Thoreau were an angel, he would no doubt hover over these lines, trumpeting his approval, not least because he revolutionized pencil-making technology in the 1840s in the family-owned pencil factory by inventing a graphite mill called a plumbago that ground graphite finer than ever before and creating a method of mixing clay with graphite that made the lead harder (Thoreau also apparently invented the number system for classifying the hardness and quality of pencils, so if you ever puzzled over the difference between a #2 or #3 pencil, you can thank him for that).
Berry goes on to explain his writing process: He composes his manuscripts in pencil, handing them off to his wife, who then types them on a 1956 Royal typewriter, making suggestions and edits along the way. He is proud of this method because it does not require any electricity, which would implicate him even more deeply in the energy corporations who were, at the time, strip mining mountaintops for coal in his home state of Kentucky. He goes on to complain that computers are costly, and he does not believe that they will make him a better writer.
This last note of skepticism is ringing in my ears lately. Will computers make me a better writer? The question is more relevant than ever in the age of generative AI. As a writer and a teacher of writing, I have watched the proliferation of chatbots with both trepidation and fascination. The promises of a digital tool that helps you write faster and more efficiently are familiar—these are the great mantras of the machine age—and yet this new wave of writing-assistive technology is different from anything that has come before it. From the first clay cuneiform tablets and the invention of papyrus, writing and technology have tangoed together across the last 5,000 years, creating the conditions for civilization itself. Writing supercharged the human gift for communication and expression, allowing ideas to straddle the earth and retain their integrity across centuries, facilitating everything from grain storage in the Sumerian empire to building and launching the James Webb Space Telescope.
The modern world is made from writing. At the start of my writing classes, I gesture out the window to the campus and recite all the ways in which writing has literally created the college—the proposals written for new buildings, programs, courses, and student organizations; the thousands of emails sent every day to manage the bureaucracy and infrastructure of the place; the proliferation of reports; the papers written by students; their emails back and forth to their professors; the research published by professors; and so on. My point is that writing makes the world we inhabit possible. And my campus is just one humble rock in a vast archipelago of writing-dependent institutions that stretches beyond the horizon.
Throughout this span of writing-powered civilization, humans have always been in the driver’s seat when we sit down to write. The great hidden conceit of writing—too obvious to even say out loud—is the centrality of human consciousness and will in the writing process, whether it is expressed in the singular genius of a great novelist or in the collaborative anonymity of an interim progress report. Try as they did, the poststructuralists were unable to unseat human agency from the art of writing. Most of us still believe that writing is a human skill and in its highest form, an art that can be mastered.
Generative AI is the big hack that threatens to overthrow this basic pillar of civilization. It is the first writing technology that finishes your sentences and composes entire documents from scratch in seconds. The sudden ubiquity of this technology raises an important question: what happens when more of the writing and communication necessary to power civilization is outsourced to artificial intelligence, with all its human biases already baked in, impossible to detect let alone contest? What happens to our capacities for creative expression, self-invention, argument, and the raw expression of our will across time and space when over-reliance on AI atrophies our collective writing and communication skills? What happens to the fabric of society when so much communication is being done by entities that do not possess sensory organs, a brain, emotions, or the capacity to experience anything?
We should just admit it: We don’t know what will happen if AI becomes the principle mediator for written language, just like we didn’t know what would happen when we strip mined mountaintops for coal in Appalachia or exposed millions of pre-teen brains to Instagram or began pumping tens of millions of tons of carbon dioxide into the atmosphere. Our technology obsessed society excels at not knowing (or caring to know) what happens when new technology is unleashed into the world, often by oligarchs who are intent to force their own vision of reality on everyone else. Release first, reflect later. Clean up the mess afterwards. This is how we do technology.
***
How could Berry or I or anyone alive in 1987 have predicted how the computer would grow in social significance, from a metal box that sat on your desk making calculations and “processing” words on a screen to an all-pervasive mediator of reality itself; that the word “computer,” like ‘TV”, would outgrow the metal box and attach itself to so much more; that computers would be miniaturized and inserted into cars, appliances, phones, and nearly every other device that defines modern existence; that billions of computers would be linked together and connected to the largest encyclopedia of knowledge in human history; that computers would be used to spy on us and Hoover up our data and shape our thoughts about everything.
Almost forty years later, the “computer” is a power that saturates the world, democratized into billions of smartphones and yet still wielded to its greatest effect by the wealthy, who can always purchase the most bandwidth, the fastest processing speed, and access to the most sophisticated computer hardware and software on the market. Among the most pernicious effects of this power is the illusion it creates of the possibility of an alternate reality, and all the escapist pleasure associated with it. The computer I used in 1987 was seductive in its own right, and I sometimes lost myself in its pleasures, but computers had yet to overwhelm our senses or willpower. I could turn it off when I was finished, and it would be waiting there for me to turn it back on when the need arose.
At some point in the 1990s, I noticed that “logging off” had replaced the act of “turning off” the power. Now the computer appeared to be always on, even when I was not using it, dormant but always ready to flash on at my command. This quality of low-power life was unsettling to me. The computer was taking on a quality of omnipresence. Decades later, when laptops became commonplace, people began putting swatches of tape over their cameras, a precaution against unwanted surveillance, but also a way of acknowledging the creepy panopticon discomfort we were all feeling. The tape over the camera was a totemic wall we erected in the vain hope that we could shield ourselves from the all-seeing eye.
Eventually, the one computer screen became two, and then three and four or more. The devices with screens were called computers and powered by computer chips, but there were also the devices and machines without screens that were also powered by chips. After 2000, it was increasingly difficult to distinguish between those mechanical things that were powered by computer chips and those that were not. By the 2010s, many of us felt like we were surrounded by computing power.
Or outstripped by it.
***
Wendell Berry depicts the act of writing as a solitary, intentional activity powered by a simple device, a pen or pencil, working over the ancient medium of paper. As I write these words in my journal using a Pilot G-2 pen, I can sympathize with the meditative pleasures of this method. Often the pen is moving mid-sentence as my brain is still working to finish my thought. This uncertainty over how the sentence will end tends to suspend the moment and the pleasure that comes with it, turning nearly every sentence into an act of discovery. On the other hand, the computer keyboard experience is much different, faster and less meditative. Often the keyboard-composed sentence is finished before I even realize it, and then later, I am skeptical of both its craft and its content because it flowed out of me so easily. Both writing situations share a common feature, however: I can locate my will and presence in the process. I am in the driver’s seat.
There is something else to put in the positive column for pen and pencil: The quality of resistance. However crude my handwriting is, the pen produces a flowing form of resistance as it touches the paper, an ease of forward motion that is felt in my fingers and hand as the instrument moves across the page. The keyboard is a different sort of resistance, with each letter felt on my fingertips but without the satisfaction of having touched the page itself, making this sometimes painful passage to reach the right margin, and the next line, and the next. When I write on a keyboard, I am aware of words floating off into a magic box.
And there is yet another advantage for the hand-held writing implement: I am less likely to revise the sentences I write with a pencil or pen because they were crafted with more deliberation. It is very likely that the keyboard does not increase my speed because I will invariably spend more time editing the sentences I compose with it.
Am I fetishizing the pencil? Not at all. I am trying to convey the idea that the technology we use to write matters, whether it is a bone stylus, feather pen, or typewriter. Those of us who are serious writers can evaluate, at a micro level, the pros and cons of each writing technology. There is no ideal technology-independent act of writing. To write is to engage technology, plain and simple.
Up until recently, writing technologies all shared a common feature: the tool is an instrument of touch, tactile and therefore accountable to the brain that powers it. What is the effect of a technology that can “write” without the thing that acts like a medium between your brain and the page? What happens to writing when a computer application does most of the mental labor traditionally associated with it? For example, the act of searching for the right world, weighing two or three options, and then choosing one. The labor of rewriting a sentence in order to fully convey an idea. The rearrangement of sentences within a paragraph to create a desired effect in the reader’s mind—to tell a story, or contrast two things, or present an analogy. The intimate engagement with style and tone.
What happens when we remove this kind of mental activity from the equation? Can we even call it writing anymore, or should we invent a new lexicon to describe this computer-generated activity?
***
Berry’s essay is still relevant. For one thing, he correctly draws a direct line between the computer as a writing tool and the expenditure in fossil fuel necessary to power it. The potential for computer power to destroy the environment through its massive energy requirements has only accelerated since Berry published his jeremiad with the mountaintops of Kentucky at the top of his mind. I was reminded of this when I read recently that the nation’s AI gurus travelled to Washington to discuss the growing energy demands of large language models (LLMs). For all of its techno-utopian promises of AI solving the world’s greatest problems, generative AI is already a massive energy pig with unlimited potential for girth-expansion. A recent article in The Verge reported that the energy needs of AI could equal the energy used by the Netherlands by 2027.
Energy gluttony aside, my antipathy towards this technology cuts to the definition of writing itself. I am shocked that people who have worked so diligently to make computers capable of mimicking human writing have such disregard for it. They appear to have abandoned the idea that quality can be measured by the rhetorical capacities of the human brain. The advertising claims of companies that use AI to generate content are telling in this regard: You can work faster. You can crank out more blog posts per hour than a human can. Your productivity will improve. These claims are vacuous. Speed and efficiency are values associated with factory production, not writing. There are sometimes valid reasons for writers to care about speed, as anyone who has written on a deadline will attest, but to proclaim speed as a primary value of writing is a sure sign of Philistinism. If the goal is to crank out more content into an already content-saturated digital ecosystem, then bravo, AI can help with that.
The creators of chatbots treat writing as if it is just another system ripe for hacking. Hacker logic is the mentality of doing things just to prove that you can, cracking open existing systems with no regard for the downstream consequences, and taking things because they are easy to take (and then justifying the theft by saying ‘you should have protected it better’ or ‘I did it for the greater good’). Hacker logic can justify the massive theft of copyrighted material to train an LLM. Hacker logic is the chutzpah that allows AI developers to argue for a massive expansion of the nation’s energy usage to support their projects with empty promises of making their tech sustainable in the future. Hacker logic allows otherwise intelligent people to believe that writing can and should be done without a human brain in the driver’s seat.
***
Berry laid down the marker in 1987: “I disbelieve and therefore strongly resent the assertion that I or anybody else could write better or more easily with a computer than a pencil.” By modern standards, this declaration seems hopelessly outdated. Few who write for a living now could adhere to such an extreme purity test. The modern workplace is completely dominated by computers, for good or ill. As a practical matter, the pen has been eclipsed by the computer, not because it is an inferior technology but because society has so thoroughly given in to the logic of computers. We write with computers in large part because they are networked together, which makes communicating our writing much easier than it was before, and because they are linked up to the world's greatest encyclopedia of knowledge, which certainly makes research more efficient. You can print 100 copies from a computer and or post it on Facebook for a thousand people to see instantly, but the act of writing itself, the sentence-by-sentence craft of it, seems no more aided by a keyboard than a #2 pencil.
Berry’s essay is steeped in the language of pragmatism, but as I read it, he is using the pencil as rhetorical staff for a broader message about the role of technology in our lives. In this regard, I stand with him, even today. This staff, held high in defiance, is the standard for techno-skepticism, which is the right to refuse the shiny new box with the yellow happy face sticker stuck to it, the box that forever promises to solve your problems without even a whisper of possibility that harm might also come with it.
The modern American deference to new technology is practically an article of religious faith. We are constantly being sold on the idea that new technology will lead to utopian outcomes—less friction, less labor, more free time for us to pursue our personal goals and enjoy leisure. Of course, the reality is always much less impressive. Somehow, the utopia of free time never materializes because the economy forces individuals to work more hours, or a second job, to make ends meet. And the same technology that frees up stretches of free time in our lives also serves up new seductions that sop up that time into an endless whirlwind of entertainment.
I need not look any further than ChatGPT to see the false promises of techno-utopianism. As a writing teacher, I see how this tech has already undermined academic integrity at every level of the writing process, while undercutting the ability of students to learn critical thinking skills through writing. This technology was built on a foundation of massive copyright infringement. It undermines the teaching of writing by subtly sending the message that process does not matter. It convinces people with underdeveloped writing skills that they are already highly proficient writers, and it delivers the message to aspiring writers that a machine that can finish your sentences is a valuable writing tool.
And now the AI bandwagon has inevitably arrived on college campuses calling everyone outside to jump on board or at least pledge fealty to the new high-tech future. Even in writing and composition classes, I see a softening in what was, two years ago, near-universal pushback against chatbots, which had suddenly emerged as the go-to cheating aid for America’s laziest and least engaged college students. Plagiarism was always a problem in college classrooms, but chatbots have created a new technological arms race that forces instructors increasingly into the labor-intensive role of integrity police. It is now easy to see a future in which most writing teachers abandon the rhetorical method altogether in favor of a new role as trainers of prompt engineers.
Techno-skepticism is uncomplicated. It requires of us only the willingness to activate our built-in bullshit detectors. When speaking of computers and writing, Berry wrote: “when somebody has used a computer to write work that is demonstrably better than Dante’s, and when this better is demonstrably attributable to the use of a computer, then I will speak of computers with a more respectful tone of voice, though I still will not buy one.” This seems like a reasonable challenge to any new tech: Can you prove to me that the product will be superior to that which came before?
Here is another example, from Thoreau, who expresses skepticism about the railroad that skirts the north end of Walden Pond, and railroads in general:
To make a railroad round the world available to all mankind is equivalent to grading the whole surface of the planet. Men have an indistinct notion that if they keep up this activity of joint stocks and spades long enough all will at length ride somewhere, in next to no time, and for nothing; but though a crowd rushes to the depot, and the conductor shouts “All aboard!” when the smoke is blown away and the vapor condensed, it will be perceived that a few are riding, but the rest are run over,—and it will be called, and will be, “A melancholy accident.”
This is a metaphor that provides us with a question that should be asked of any new technology: what or who will be run over in the rush to make this new thing widely available to all? It is always something, or someone.
Opmerkingen