"The Fifth Decimation" and Other Chatbot Atrocities
The first time I witnessed ChatGPT "hallucinate," I was discussing my own writing with it. As a writer and a writing teacher, I am alternately fascinated and horrified by chatbot technology. I am also keen to test its capabilities, so I recently asked ChatGPT to analyze the style of three short stories I have written, which it did very well. A+. But in responding to my prompt, the chatbot mentioned a fourth story that I had not written, something called “The Fifth Decimation.” I checked the Internet; as far as it knows, no such story exists.
When confronted with this confabulation, the chatbot promptly apologized (it does that when proven wrong, and I appreciate it). Then, feeling a bit frisky, I asked it to actually write a story called “The Fifth Decimation." The story it spat out, which was more of a plot summary than an actual story, is a dull seven-paragraph mashup of dystopian narratives about post-apocalyptic underground societies. It goes something like this: In a near future ravaged by climate change and war, an underground community periodically expels a fifth of its population to the surface with meager supplies. As the next culling approaches, there is a rebellion led by a hacker named Maya. The rebels get access to resources hoarded by "The Council" controlling this underground city and then flee to the surface where they rebuild society.
I am fascinated that a computer application can so quickly spin up a story based on a title, even if it is derivative, unimaginative, and frankly unfit for human consumption in any capacity other than to be the object of mockery or curiosity. My fascination quickly turns to discomfort, however, when I begin to imagine how generative AI might alter societal attitudes about creativity, especially among young people who grow up using it as a ubiquitous tool. Generative AI applications are black boxes. Their creative processes are invisible to us. Will widespread use of these tools erode users' expectations of the creative process? After all, AI can very quickly generate slick, finished, consumable creative work. Will this basic functionality of the tool subtly train its users to expect the creative act to be instantaneous and struggle-free?
This is the thought that keeps me up at night.
For the past eight months, I've been bombarded with orgasmic praise for AI's creative capabilities. AI can write a sonnet about Captain Kirk, compose a song about butterflies in the style of Rage Against the Machine, or spit out a photorealistic image of a blueberry muffin shaped like a dog's head. This praise often revolves around the AI's technical ability and capacity for stylistic mimicry. In most human creatives, technical ability and craft are cultivated through many hours of practice working with various mediums, techniques, materials, and built-in limitations. Will future creatives care less about devoting the time and effort to learning these skills? Pablo Picasso worked fourteen hours a day at his craft, but why should anyone work so hard to be Picasso when the AI can so quickly and effortlessly create an image in his style? This is the cynical question looming in the shadows every time a human prompts an AI to engage in a creative act.
And speed. Everyone is gobsmacked over the speed at which these creative acts occur. On Facebook, I am now bombarded with advertisements for AI writing services that promise to write blog posts "10X faster." I wonder, is there really a demand for 10 times as many blog posts on the internet, which is already littered with many thousands of dead, unread blogs? One of the more well-known of these applications, Sudowrite, gushes about how you can now "write your novel or screenplay faster" than before. I am amused by this advertising pitch because speed is almost never a primary value for serious writers, except for those of us who work in deadline-intensive workplaces, and in these situations, we are always fretting about the tradeoff between deadlines and quality.
This uncritical salivating over speed and efficiency has a familiar ring. This new wave of generative AI is consistent with decades of societal precedent in which technology is glorified for its efficiency, time-saving, and cost-effectiveness. The veggie chopper that chops, spirals, and juliennes instantly; the quick-drying nail coat that dries in just two minutes; the mini steamer that eliminates the need for ironing; the microwave rice maker, done in just minutes; robot vacuums, air fryers, and instant pots; and the "Wet and Forget" shower cleaner. The new generative AI promises that writing can now be judged by the same standards we apply to many of our household appliances.
Art has always been about resistance. The mind conceptualizes something—a musical composition, an image, a story—and the artist struggles to manifest it in the world. The struggle sometimes involves the difficulty of working within a particular medium with available tools and materials that are not perfectly suited to the task. Many creatives will tell you that it is limitations that often spark a flourishing of creative activity. Sometimes artists grapple with their own insecurities and perfectionism as they initially try and fail to manifest their vision in the work. For writers, the struggle is often with words themselves, which are imperfect mediums for thoughts and feelings.
But process is not merely a series of steps one takes in the act of creating something; it is itself generative. The countless hours spent standing at an easel will create new pathways in the brain, and any artist will tell you that the physical labor of making art changes the body itself. The sculptor's hands are different because of her work. In his book This is Your Brain on Music, neuroscientist and musician Daniel Levitin explains how making music improves memory, mood, and cognitive functioning. I am certain that many hours of writing have changed my brain. In a study led by Martin Lotze of the University of Greifswald in Germany, researchers observed that brain scans of experienced writers looked different from those with much less experience. But most creatives already know that we are being made and remade by the struggle to create.
For me, the scariest aspect of generative AI is how efficiently it narrows the scope of choice within any creative process. Another of the taglines for Sudowrite hints at this phenomenon by promising to be "autocorrect on steroids." We should all be terrified by the implications of this promise. Autocorrect essentially uses AI to suggest just a few possible completions to a sentence you have already begun, thereby greatly reducing the vast field of possibility for a thought to become a sentence, while also encouraging a dependency in users for digital assistance in completing their thoughts. It infantilizes us. It turns over the act of thinking for ourselves to computer applications created by corporations. What will happen to our society when the underlying logic of "autocorrect on steroids" is applied to all aspects of human life? In what other areas of our lives will we surrender agency and autonomy to thinking machines?
In the short term, however, I worry most about the looming tsunami of AI garbage that will soon flood every portal of human communication. The wave has already begun to break. Literary magazines are now being deluged with crappy AI-generated short stories submitted by dullards who have no appreciation for the skill, craft, and time required to write an excellent narrative. Are these budding prompt engineers now to be considered "writers"? Many of them no doubt aspire to this role. After reading "The Fifth Decimation," I now have sympathy for the magazine editors around the world whose inboxes are being doused by this new high-flow firehose of dreck. For them, the great AI Crapocalypse has already arrived.