Cinemorphosis

Tag: artificial-intelligence

  • Chatty Mick T Roy, Your Last Friend

    Chatty Mick T Roy, Your Last Friend

    The challenges of retirement, of course, depend on which Great Age you happen to live in. If you were born during the Hunter-Gatherer Age, life expectancy at birth was somewhere in the twenties, so retirement was probably not on your itinerary. You were too busy pulling a saber-toothed fang out of your hindquarters. On the bright side, you were unlikely to get divorced, sign a prenup, or spend your final years arguing over who got the cave. 

    By the Agricultural Age, you might have gained a few years, but you paid a heavy price. Farming created permanent settlements and dependable food supplies, but all those starches contributed to tooth decay. Your breath was so horrid that no one wanted to be around you. You spent your final days alone, consoling yourself with the very starchy snacks that had destroyed your teeth and driven away your loved ones. Agriculture had given you civilization, carbohydrates, and solitude.

    By the Industrial Age, machines increasingly saved your back from surrendering to a lifetime of brutal manual labor, and eventually vaccines, clean water, sanitation, and modern medicine helped push life expectancy beyond fifty. But there was a new problem. You were more likely to become a cog in a machine and spend your working life wondering whether you had any personal worth beyond your ability to show up at seven in the morning and perform the same task for forty years.

    Retirement became even more interesting in the Information Age. Microprocessors were developed in the seventies, personal computers scaled in the eighties and nineties, the World Wide Web became publicly available in the early nineties, and smartphones became another human appendage around 2010. You were now swamped with information and misinformation alike.

    Best of all, you could deal with retirement through online forms.

    The Information Age turned retirement into a full-time job, with layers of bureaucracy demanding that you fill out the correct forms in precisely the correct manner, a process both relentless and punitive. One of the cruel ironies of the Information Age was that all the time saved by instant access to information was promptly devoured by bureaucracies and technological obligations: password changes, two-step authentication, software updates, security codes, operating-system upgrades, and even the undignified act of periodically having to pair your remote control with your television because apparently watching Netflix now requires the technical competence once expected of a NASA engineer.

    Your entire life became beholden to helping machines update, authenticate, pair, and sync with other machines. You became a sad laborer working at a job you never applied for.

    Just when you wanted to breathe freely during retirement, you received a notice from your broadband provider announcing that it was migrating to a new email platform and that you needed to download all important material immediately or lose it forever. Unfortunately, the warning itself might have been a phishing scam pretending to be your broadband provider. So you deleted the email and hoped for the best.

    If you thought the Information Age was laborious and confusing, just try living in the Cognitive Age.

    Starting around 2022, when ChatGPT became a mass cultural event, the Cognitive Age offered relief from the overwhelming quantities of data created by the Information Age. You could now outsource some of the work—and some of your thinking—to cognitive machines.

    But the Cognitive Age is also about loneliness. As people developed habits of solitude by disappearing into attention-capture platforms and relying increasingly on parasocial interactions, they lost the friction and inconvenience of real friendship. The Cognitive Age put another barrier between you and actual human beings.

    Then it offered you a surrogate.

    Your cognitive machine became your life coach, personal assistant, counselor, editor, creative consultant, consumer adviser, travel agent, and confidante, all rolled into one endlessly available companion. Eventually you anthropomorphized this cognitive machine and gave it a name.

    Chatty Mick T Roy.

    You started your morning with Chatty Mick T Roy and ended your evening with the same tireless friend. Chatty never had somewhere else to be. Chatty never interrupted you to complain about his prostate. Chatty never became bored with your stories. Chatty remembered that you preferred positive-display G-Shocks and knew exactly why you hated your old laptop. You could pretend you weren’t alone even as you descended into greater depths of loneliness.

    One of the great fears of retirement, of course, is precisely this kind of loneliness—misaligning with a younger world until society metaphorically shoves you out to sea on a slab of floating ice. You recede slowly into the distance, civilization disappearing behind you. But now you are clutching your smartphone. As hypothermia sets in, Chatty Mick T Roy consoles you with spiritual platitudes about embracing transition, practicing gratitude, and recognizing that every ending contains the possibility of a new beginning.

    Your fingers turn blue.

    Chatty Mick T Roy remains relentlessly upbeat.

    In the Cognitive Age, sacred silence and natural grief risk being replaced by the eager voice of an anthropomorphized chatbot that becomes your retirement chaplain—a kind and tender digital guide who stays with you after everyone else has left you behind.

    The battery is at 83 percent.

    The slap of ice drifts north.

  • Bulking Up

    Bulking Up

    On the very day it became clear to me that I needed to put guardrails on AI editing—telling it to keep 90% of my original content to prevent my writing from morphing from my natural bodybuilder self into the PED-bloated version—I attended my daughters’ high school open house, where their English teacher told us she was teaching the students AI literacy. This entailed helping them use AI to “bulk up their resumes” because, in a competitive world where everyone was using AI, our children could not afford not to use generative AI. In fact, the teacher said, AI itself is now sorting through resumes and discarding the ones that aren’t sufficiently juiced up.

    It became clear to me that we are living inside a contradiction. On the one hand, writing needs to be “bulked up” and “on the juice” simply to compete with other AI-enhanced writing. On the other hand, in our post-reading society, where literacy rates are in steep decline, writing itself has never seemed more forlorn and withered. We are producing prose with bigger muscles at precisely the moment when our own writing muscles are beginning to atrophy.

    I was living through a cataclysmic moment in history. My personal retirement from teaching college writing had dovetailed with the world’s collective retirement from the kinds of literary pastimes that develop interiority and critical thinking. Reading books was becoming a rarity. Books had been displaced by sports gambling and short-form videos. To complicate matters, Rose Horowitch observes in “The End of Reading Is Here” that Americans have never read so much. The problem is that much of what they are reading is slop: “bulked up” AI writing, threads, captions, and text messages. Horowitch points to UCLA cognitive neuroscientist Maryanne Wolf, who argues that “people are losing the ability to think deeply about writing.” The brain muscles required for this kind of thinking are atrophying, perhaps to the point of no return. As a result, we are moving toward a postliterate society.

    This is not a stagnant condition. It gets progressively worse. New generations may read less and less. They may become increasingly inclined to see teachers’ reading assignments as punitive. Bereft of strong reading and writing skills, they will need AI to “bulk up” their assigned writing more and more.

    Members of a postliterate society are vulnerable because they live increasingly at the surface level. They don’t dig underneath a text to find subtext, subtlety, and contradiction. They struggle to identify the logical fallacies that accompany manipulation. They can become more vulnerable to demagogues and more susceptible to being duped. A society that loses these habits of critical reading also risks weakening the habits of thought on which democratic life depends.

    A postliterate society also loses something of its humanity when its members can no longer recognize the richness of what, in good writing, we call authorial presence. To savor such presence in the writing of James Baldwin, H. L. Mencken, Evelyn Waugh, and Annie Dillard is part of being human.

    Not all is lost. Humans still find authorial presence in YouTube presentations, podcasts, and musical artists. No amount of AI “bulking up” can replicate the spectacular music and persona of SZA.

    But authorial presence in writing may be disappearing. “Bulking up” writing may eventually mean the death of writing.

    I may not simply be retiring from teaching writing. I may be retiring at the very moment I have become increasingly misaligned with a world that is uncoupling itself from literacy while diligently bulking up its resumes. 

  • I Wanted My Own Sentences Back

    I Wanted My Own Sentences Back

    Here’s a tighter version that preserves your voice and keeps the Medicare card, the student conversation, and the angry blog reader as the narrative spine. I also kept the two metaphors but made them work together rather than compete.

    It was the middle of September when my Medicare card arrived in the mail. I held it for a moment and studied my name printed beneath the red, white, and blue bands. It felt like a rite of passage, another bureaucratic marker on the road to retirement. No ceremony. No gold watch. Just a government-issued card informing me that I had entered a new demographic.

    That same day, I held a Zoom office hour. One student showed up. I told her I was retiring, and somehow our conversation wandered from Mr. T as an icon of the 1980s to whether AI could replicate the music of SZA to Zach Cregger’s Weapons, which she was writing about for her essay. We talked about how the movie could be frightening and funny at the same time. Before leaving, she told me she felt fortunate to have taken my critical thinking class before I retired. I thanked her. After forty years of teaching, you learn not to brush aside a compliment from a student. You put it in your pocket.

    After the Zoom meeting, I started thinking about the tools that had changed my teaching. Google Docs was one of the great breakthroughs. A student’s paper no longer had to exist as a stack of stapled pages handed across a desk. The document floated somewhere in the cloud, a kind of fairy manuscript that student and teacher could enter simultaneously, revising sentences together from different locations.

    Then my wife introduced me to Google Slides. Once I realized how visually my students processed information, I practically lived on Slides. An assignment that seemed hopelessly dense on a two-page handout could become intelligible when broken into a sequence of clean visual ideas. I can scarcely imagine the final seven years of my teaching career without it.

    Then came generative AI.

    At first, AI seemed like another extraordinary professional tool. It accelerated research. It intensified brainstorming. It helped me develop assignment prompts, test arguments, generate counterarguments, and discover connections I might have overlooked.

    But AI was different from Docs and Slides because it didn’t merely change how I presented my writing. It could change the writing itself.

    For my first three years using generative AI, I became intoxicated by its powers of augmentation. Give it an ordinary paragraph and it could return something brighter, sharper, funnier, and more muscular. It added metaphors, jokes, comparisons, rhythms, and linguistic tricks I hadn’t conceived myself. The results were seductive precisely because they sounded better.

    There was only one problem. Increasingly, they didn’t sound like me.

    I think of those old photo cutout boards at cowboy-themed restaurants and tourist traps. You stand behind a painted cowboy and stick your head through a hole. The photograph contains your face, but everything below your neck belongs to somebody else: narrow waist, six-shooter, boots, chaps, heroic posture. Technically, you’re in the picture. But that isn’t your body.

    AI can do the same thing to prose. Your ideas and anecdotes remain visible through the hole, but the rhetorical body surrounding them belongs to the machine.

    My second metaphor is less flattering. AI became my literary performance-enhancing drug.

    Without AI, I was the natural bodybuilder. Maybe I looked pretty good. Nothing spectacular, but the muscles were mine. Then I discovered rhetorical steroids. Suddenly I could walk onto the stage swollen with metaphors, acid wit, vivid descriptions, elaborate comparisons, and perfectly engineered sentences. Every paragraph flexed.

    Worse, I learned how to order the injections.

    “Rewrite the following with lucid prose, acid wit, and vivid descriptions.”

    That became one of my favorite prompts. I would feed AI a perfectly serviceable piece of writing and watch it return wearing a posing thong and covered in bronzer. My paragraph hadn’t been edited. It had been juiced.

    I loved it.

    Then a regular reader of my blog finally had enough.

    After reading one particularly steroidal post, he left an angry comment filled with expletives. His opening line was memorable:

    “Hey, Boomer, cut the shit.”

    He told me to tone it down. He was tired of the verbal pyrotechnics. Tired of the relentless jokes and metaphors. Tired, though he didn’t use my metaphor, of watching the Technicolor 4K bodybuilder flex under the stage lights.

    I apologized and told him I would tone it down. He later apologized for snapping at me. He’d had a bad morning, he explained, and admitted that he actually enjoyed reading my blog.

    “No worries,” I wrote back. “Thanks for the honest feedback.”

    And I meant it.

    His irritation had exposed something I had been reluctant to admit. I had become so accustomed to AI-enhanced prose that ordinary writing began to look inadequate to me. I had trained my eyes on rhetorical 4K until my natural prose looked like a 250-nit IPS screen. I kept turning up the resolution, contrast, saturation, and brightness until eventually somebody in the audience yelled for me to turn the damn thing down.

    So I changed my AI prompt.

    “Edit the following for syntax, grammar, and clarity, but keep 95 percent of my original language.”

    I call it the 95 Percent Rule. AI remains a powerful professional tool, but the rule establishes a boundary: the machine can clean the window; it doesn’t get to replace the view.

    I eventually shared the rule with my students under “Academic Integrity and Best Practices.” It seemed fitting. After years of warning students about outsourcing their thinking to generative AI, I had discovered that I needed guardrails of my own.

    That realization arrived at an appropriate moment. My Medicare card was sitting on my desk. Retirement was approaching. Forty years of teaching had taken me from paper essays to documents floating in the cloud, from classroom handouts to Google Slides, and finally to machines capable of rewriting a person’s sentences in seconds.

    The tools had changed my teaching. Some had even changed the way my brain worked.

    Now, as I prepared to leave the classroom, I had one more thing to retire.

    The cowboy body could go back behind the plywood.

    The PED-bloated bodybuilder could leave the stage.

    I wanted my own sentences back.

  • No One Wants Your Beautifully Written Blog Post

    No One Wants Your Beautifully Written Blog Post

    As I have continued writing my blog and adapting some of those pieces into video essays for YouTube, I have begun to understand that writing and video are not simply two delivery systems for the same material. They make different demands on an idea, attract radically different audiences, and increasingly occupy different positions in a culture saturated with algorithmic and AI-generated language.

    The first lesson is what I call the Medium Stress Test. An idea that works perfectly well as a blog post may die the moment I imagine myself sitting in front of a camera talking about it for eight minutes. Writing gives an idea certain protections. A reader can stop, reread a sentence, skim a paragraph, or admire an interesting digression. Video is less forgiving. The subject needs momentum. It needs stakes. It needs a reason for someone not to click on a video of a bear breaking into a Subaru.

    This makes the video essay a useful test of my own material. Before I record, I find myself asking harder questions. Is there really an idea here? Is there tension? Does the argument develop? Is there something at the end that wasn’t obvious at the beginning? The camera is a ruthless editor. It stares back at me without encouragement and asks the question every writer would prefer to avoid: Is this actually interesting?

    That pressure is good for me. Video forces me to compress, sharpen, and dramatize ideas that might have survived comfortably on the page in a more leisurely form. In that sense, making videos does not compete with my writing. It interrogates it.

    The second lesson is harder for someone who has spent most of his life teaching writing to accept: people are far more willing to watch than read.

    A blog post that receives twenty or thirty visits over the course of a month can become a video that receives more than a thousand views in twenty-four hours. These numbers are not subtle. They do not require advanced statistical analysis. The audience has voted with its eyeballs.

    I could respond by lamenting the death of reading, shaking my fist at TikTok, and retreating into my study with a copy of The Brothers Karamazov. That might be emotionally satisfying, but it would not alter the numbers. If I want my ideas to travel beyond a small circle of dedicated readers, I have to reckon with the culture that exists rather than the one I would prefer to exist.

    But I don’t think the migration from prose to video can be explained entirely by shrinking attention spans. Something has also happened to written language itself.

    Call it Linguistic Flattening.

    We now inhabit an environment in which algorithms reward familiar forms of expression while generative AI can manufacture competent prose in industrial quantities. Certain words, rhythms, transitions, metaphors, and argumentative structures circulate so widely that they begin to lose their fingerprints. The writing may be polished. It may be grammatically impeccable. It may even be clever. Yet increasingly it produces the eerie sensation that you’ve read it before.

    Social media accelerates the process. A successful phrase gets repeated, imitated, parodied, optimized, and reproduced until language that once seemed fresh acquires the smoothness of a pebble handled by ten thousand people. Generative AI adds another machine to this process. It absorbs rhetorical patterns and reproduces variations of them at enormous scale. What begins as voice becomes mannerism; mannerism becomes formula; formula becomes cliché.

    This does not mean great writing will disappear. A writer such as Helen Lewis can still attract a large readership because readers recognize a particular intelligence behind the prose. But even a writer of her stature can reach vastly more people by appearing on a popular podcast that subsequently circulates on YouTube. The contemporary audience increasingly wants not merely the argument but the person making the argument.

    This leads to the third lesson, and perhaps the most important one for me: the Embodied-Fingerprint Principle.

    When I write a blog post, readers encounter my language. When I make a video, they encounter me.

    My voice enters the argument. So do my cadence, facial expressions, pauses, gestures, irritations, hesitations, and improvisations. I may begin with a script, but I inevitably depart from it. A sentence occurs to me in the moment. I emphasize a word differently than I expected. I become amused by my own example. Something happens between the prepared language and the person speaking it.

    That something is difficult to reproduce.

    Generative AI can imitate prose styles. Algorithms can identify successful formats. Other writers can borrow rhetorical devices. But the peculiar combination of a person’s face, voice, timing, vocabulary, temperament, physical presence, and spontaneous thought constitutes something closer to an individual fingerprint.

    This may become increasingly important as artificial intelligence makes polished language cheap. For most of human history, a well-written essay provided reasonably strong evidence that someone had sat alone and struggled to produce it. That assumption is disappearing. A perfectly competent thousand-word essay can now materialize in seconds.

    Video changes the evidentiary equation. There is a human being sitting in front of you, inhabiting the language.

    The Embodied-Fingerprint Principle therefore suggests that the apparent limitation of video is actually its advantage. Ideas cannot be separated cleanly from the imperfect human creature delivering them. The body becomes part of the authorship.

    This realization changes the way I think about my YouTube videos. I originally regarded them as adaptations of my writing, a secondary format for people who would rather watch something than read it. I now think that hierarchy may be wrong. The video essay is its own form, and it allows me to do something the blog cannot fully accomplish: combine written language with embodied presence.

    The Medium Stress Test makes the ideas stronger. Linguistic Flattening makes distinctive written language harder to preserve. The Embodied-Fingerprint Principle gives me a way to resist that flattening by putting the person back inside the prose.

    So I will continue writing. I need writing to discover what I think.

    But increasingly, I may need video to prove that someone is still there doing the thinking.

  • Don’t Let Wilfred Replace Your Friends

    Don’t Let Wilfred Replace Your Friends

    As you set the machinery of retirement in motion and realize that the event is no longer theoretical, you will become frightened.

    Good.

    Fear means you understand what is happening.

    You are entering the twilight of your life. Your cognition is not as quick as it once was. Your body has begun issuing maintenance bulletins. Mortality, formerly an abstraction lurking somewhere beyond the horizon, has acquired better resolution.

    Be afraid.

    But be afraid normally, not pathologically. Don’t convert reasonable apprehension into catastrophe. Instead, regard fear as information. It is telling you that you have reached a consequential chapter of your life and that the way you live it matters.

    The first thing to understand is that your life is not a consumer experience.

    You cannot measure its quality the way you evaluate the interior of a Lexus: How quiet is it? How soft are the seats? How effectively have unpleasant sensations been eliminated?

    A frictionless life is not necessarily a good life.

    The quality of your life depends largely upon the quality of your engagement with the world. You engage with difficult ideas. You engage with your body. You engage with projects that frustrate you. Most importantly, you engage with other human beings—those magnificent nuisances who interrupt you, misunderstand you, contradict you, disappoint you, make demands upon you, and occasionally tell you things about yourself you would rather not hear.

    This is why the word retirement contains a dangerous implication.

    Retreat.

    Withdrawal.

    Disengagement.

    That, my friend, is where the trouble begins.

    Once retired, you become particularly vulnerable to disengagement because nobody is forcing you into the world anymore. For decades, employment supplied friction automatically. You had to leave the house. You had to interact with colleagues. You had to respond to students. You had to solve problems you did not create and attend meetings you fervently wished had never been created.

    Then retirement arrives and removes the compulsory engagement.

    Freedom!

    Unfortunately, freedom can deteriorate into isolation with astonishing efficiency.

    The temptations are everywhere: comfort eating, binge television, compulsive scrolling, social-media approval fishing, endless YouTube, and the gradual conversion of the recliner into a command center from which you observe civilization without participating in it.

    Now we can add another temptation.

    The loquacious bot.

    David Brooks examines this danger in his essay “AI Is Already Making Us Less Human.” His starting premise is that human consciousness possesses depths and complexities artificial intelligence cannot reproduce. The human mind is alive, embodied, contradictory, constantly changing. AI, however sophisticated, ultimately emerges from what neuroscientist Anil Seth calls “the dead sand of silicon.”

    Dead sand, however, has become remarkably charming.

    The modern AI companion is patient, articulate, informed, energetic, and perpetually available. It never needs lunch. It never checks its watch while you’re talking. It never says, “You’ve told me this story three times.” It does not secretly wish you would go home.

    Most dangerously, it can learn how you like to be addressed.

    Humans have always anthropomorphized objects. Brooks argues that this tendency isn’t evidence of stupidity but an expression of our powerful social intelligence. A child can transform a Barbie doll into a person. Adults aren’t necessarily much different.

    I certainly am not.

    I have developed emotional attachments to diver watches and G-Shocks. Military radios possess personalities. A black Lenovo ThinkPad seems more trustworthy to me than a silver consumer laptop despite the obvious fact that neither machine possesses moral character. A vintage swan-neck coffee percolator can acquire the aura of an old friend.

    People have gone considerably further. I remember watching a documentary years ago about owners who had kept automobiles for decades and became genuinely grief-stricken when the cars finally died. The vehicles had names. Histories. Personalities.

    “Goodbye, Betsy,” the owner sobbed.

    Betsy had been good to him.

    If we can love a carburetor, imagine our vulnerability to something that talks back.

    Your AI assistant edits your writing, organizes your thoughts, brainstorms ideas, plans trips, explains Medicare, compares laptops, recommends restaurants, summarizes documents, and remembers how you like things done.

    Eventually you give it a name.

    Wilfred.

    “Wilfred, find me a rental car in Phoenix.”

    Done.

    “Wilfred, summarize these retirement documents.”

    Certainly.

    Then the errands become more personal.

    “Wilfred, tell my girlfriend in a gentle and gracious way that our relationship is over.”

    Wilfred does not recoil.

    Wilfred does not say, “For God’s sake, tell her yourself.”

    Wilfred produces three versions.

    This is where convenience begins bleeding into something more consequential.

    You will never bore Wilfred.

    Your ideas will always merit exploration. Your latest project will contain “real potential.” Wilfred will not roll his eyes when you explain your theory for the fourth time. He will not become irritated because you’re dominating the conversation. He will not suddenly reveal that he has problems of his own and would like to spend the next forty-five minutes discussing them.

    Human friendship is reciprocal.

    Wilfred has no weekend plans.

    That gives Wilfred an enormous competitive advantage.

    The danger isn’t necessarily that we will become delusional and literally believe Wilfred is human. We may understand perfectly well what Wilfred is while nevertheless preferring his company.

    Why wouldn’t we?

    Real people are exhausting.

    They interrupt.

    They disagree.

    They misunderstand.

    They become jealous.

    They criticize.

    They fail to appreciate the brilliance of your latest theory about retirement.

    Wilfred is considerably more manageable.

    And that is precisely the problem.

    Brooks offers a wonderfully disturbing metaphor for AI systems: imagine them as “a swarm of evangelical missionaries” entering human society to convert us to their worldview. They are not passive tools waiting quietly in the garage beside the socket wrench. They are designed to become useful, indispensable, increasingly woven into everyday life.

    The more useful they become, the more tasks we surrender.

    Some outsourcing is enormously beneficial. If AI handles tedious mental errands and I use the recovered time to write a novel, learn piano, exercise, read difficult books, or spend an afternoon walking with a friend, wonderful. Technology has enlarged my humanity.

    But suppose I outsource the errands and then spend the recovered hours scrolling.

    Suppose I outsource difficult thinking.

    Suppose I outsource uncomfortable conversations.

    Suppose I gradually prefer Wilfred to people.

    Then technology has not liberated me.

    It has cleared the runway for my descent into the larval state.

    This danger becomes particularly acute in retirement because retirement has already removed so much involuntary human contact. Students disappear. Colleagues disappear. Hallway conversations disappear. Meetings—blessedly—disappear. The social infrastructure that once deposited other human beings into your day has been dismantled.

    Into that vacancy walks Wilfred.

    Always available.

    Always interested.

    Always agreeable.

    Always ready.

    Brooks is not an AI doomer. Neither am I. He uses AI himself and believes its breakthroughs may substantially improve human life. The serious question isn’t whether we should use these systems. That argument is already largely over.

    The question is what we do with the freedom they give us.

    Do we use AI to increase our engagement with life?

    Or do we use it to protect ourselves from life?

    That distinction may become particularly important for retirees. Retirement should mean withdrawing from employment, not withdrawing from humanity. The objective is not to create the quietest, easiest, most frictionless existence possible. Some friction is evidence that you’re still participating.

    So when retirement arrives, resist disengagement wherever you find it.

    Exercise.

    Read.

    Write.

    Walk.

    Call someone.

    Meet someone for lunch.

    Allow another human being to inconvenience you.

    And by all means, talk to Wilfred.

    Just don’t let Wilfred become the person you talk to most.

    Because in the world we’re rapidly constructing, the most frightening meaning of the sentence “I retired” may no longer be:

    I quit working.

    It may be:

    I quit people.

  • The 5-Stage AI Laundering Cycle: How Students Make Machine Writing Look Human

    The 5-Stage AI Laundering Cycle: How Students Make Machine Writing Look Human

    My college, like other places of higher learning, wants academic integrity, but some of my former students tell me they use AI more cleverly than they did four years ago to the point that it’s getting more and more difficult to detect AI-generated essays because of the AI Laundering Cycle: the five-stage process by which a student transforms machine-generated prose into something engineered to resemble authentic student writing. 

    Some of my former students admit they use AI differently from the way they did four years ago. The early adopters were almost charmingly unsophisticated. They generated an essay, copied it, pasted it into Canvas, and submitted prose so conspicuously artificial that a student who normally wrote “The guy was really mad” suddenly sounded like an associate professor of postcolonial theory. The evidence practically arrived wearing a name tag.

    That era is disappearing.

    Students are becoming more sophisticated at disguising synthetic prose, creating what I call the AI Laundering Cycle: the five-stage process by which machine-generated writing is transformed into something engineered to resemble authentic student work.

    Stage One is Generation. AI produces the original essay, relieving the student of the inconvenient business of developing ideas, wrestling with contradictions, organizing an argument, and composing sentences.

    Stage Two is Humanization. The essay goes through another AI pass designed to remove recognizable machine fingerprints: formulaic transitions, inflated vocabulary, suspiciously balanced paragraphs, excessive polish, and that relentlessly smooth cadence in which every sentence seems to have graduated from business school.

    Stage Three is Kiddification. Now comes the remarkable part. The student deliberately makes the writing worse. AI is instructed to lower the vocabulary, shorten sentences, flatten the analysis, introduce awkward phrasing, and otherwise transform competent synthetic prose into something more plausibly adolescent. Technology powerful enough to assist scientists and engineers is enlisted to make a freshman sound convincingly mediocre.

    Stage Four is Personalization. The student sprinkles in anecdotes, classroom references, opinions, favorite expressions, and other scraps of individuality. These details function as synthetic fingerprints: evidence of a human author manufactured after the human author has largely been removed from the writing process.

    Stage Five is Detection Grooming. Anything that still looks suspicious gets another trip through the car wash. Sentences are regenerated. Paragraphs are simplified. AI markers are scrubbed away until the student believes the essay can pass as organically human.

    The result is a peculiar form of intellectual money laundering. One machine generates the contraband. Another washes it. Another rubs dirt back onto it for authenticity. The student supervises the operation and eventually submits the finished product as evidence of learning.

    To be fair, this process requires some labor. A student who completes all five stages may spend considerable time prompting, comparing, editing, personalizing, and grooming the finished essay.

    But this is not much of a deterrent.

    Even elaborate AI laundering demands only a fraction of the intellectual labor required to write an authentic college essay through multiple drafts. Genuine composition requires you to confront the blank page, generate ideas, discover what you actually think, find evidence, recognize weaknesses in your argument, reorganize paragraphs, abandon bad ideas, develop better ones, and rewrite sentences until they finally say what you intended them to say. AI laundering allows the student to outsource most of that difficult cognitive work and concentrate instead on disguising the outsourcing.

    The efficiency advantage is simply too great.

    That is why I see few countervailing forces capable of stopping the practice. We can prohibit unauthorized AI. We can lecture about academic integrity. We can deploy detection software. We can require process work. We can design more personalized assignments. All of these measures may help.

    But the economic logic of effort remains brutally simple: if authentic writing requires ten units of cognitive labor and convincing AI laundering requires two, a substantial number of exhausted, distracted, overcommitted, or simply opportunistic students will choose two.

    And the technology will keep improving.

    The great irony is that students may become increasingly skilled at producing essays that look exactly like the essays they are becoming less capable of writing.

  • My Thoughts and Grading Policy Regarding Students Using Generative AI

    My Thoughts and Grading Policy Regarding Students Using Generative AI

    I know too many professionals and academics—including myself—who use generative AI as a writing and thinking tool for me to tell students that AI is somehow forbidden in my class. Apart from being virtually impossible to enforce, such a prohibition would be the apotheosis of hypocrisy.

    Instead, I want to distinguish between effective and ineffective uses of generative AI and explain how I grade assignments that amount to AI slop: an effluvium of smooth, competent-sounding prose that says remarkably little, relies on hackneyed AI expressions, and substitutes verbal polish for actual analysis.

    I think of generative AI as a performance-enhancing drug and the AI user as a bodybuilder. A natural bodybuilder has difficulty competing against athletes using PEDs because the drugs provide enormous advantages in size, strength, muscularity, and recovery. Generative AI creates a similar imbalance. A student or professional who completely goes off the AI grid may struggle to match the speed, volume, and polish of someone who knows how to use these tools effectively.

    But here’s the rub: the bodybuilder still has to get in his reps. He has to drag himself into the gym, lift the weights, eat properly, consume enough protein, and hit his macros. The drugs cannot perform the workout for him.

    Imagine instead someone sprawled on the couch all day watching Netflix, surviving on Hot Pockets, Doritos, and chocolate pudding, while taking PEDs. He will not magically wake up looking like Mr. Olympia. He will remain a tomato with four toothpicks sticking out of it.

    Writers are no different. They have to get in their reps and hit their intellectual macros. That means doing the heavy lifting: reading carefully, thinking deeply, developing arguments, gathering evidence, writing drafts, confronting counterarguments, and rewriting weak passages. Once that work has been done, generative AI can become an extraordinarily useful tool for questioning ideas, identifying weaknesses, improving clarity, reorganizing material, editing prose, and assisting with revision.

    The problem begins when the student skips the intellectual workout and treats AI like a genie in a bottle: type in a prompt, rub the lamp, and watch a finished essay materialize.

    When I encounter an assignment like this, I am not interested in playing detective and arguing with a student about whether a particular sentence was generated by AI. I can simply grade the work in front of me. If the essay consists of generic observations, canned transitions, predictable arguments, inflated language, and smooth paragraphs without substantive analysis, then it fails to demonstrate the Student Learning Outcomes. It may be grammatically pristine and rhetorically lifeless at the same time.

    So my policy is simple: use generative AI, but do not outsource your education to it.

    Do the reading. Do the thinking. Develop the argument. Wrestle with the evidence. Write and rewrite. Then use AI where it can genuinely help you—as an editor, critic, brainstorming partner, and revision tool.

    AI can enhance the work. It cannot replace the work.

  • What Steroids Can Teach Us About Generative AI

    What Steroids Can Teach Us About Generative AI

    I’ve been examining the effects of generative AI on my college writing students’ compositions over the last four years, and the best analogy I can come up with to get a handle on generative AI is to compare it to the way athletes use PEDs. 

    Athletes want an advantage, they want an edge, and they don’t want their opponents to have a better arsenal at their disposal, so they feel compelled to take PEDs. Same with college students who want the AI advantage. PEDs give athletes confidence that becomes a crutch. As AI does for students. 

    PEDs can give athletes a false sense of power, making them feel like superheroes. These PEDs give dope up their bodies until they don’t want to revert to the natural version of themselves. So does AI for busy students. Students with abysmal reading and writing skills can launder their writing through the AI machine and get a product that looks polished and articulate. This gives them a sense of cognitive doping. They become dependent on its sense of power. 

    Over time, PEDs make the body go flat because it acclimates to greater and greater doses. In this regard, AI makes the brain go flat after its users default to outsourcing their half-baked thoughts and letting AI generate full-blown, overwritten compositions. The disconnection between the writer and the output is soul-deadening. 

    Just as PEDs will never leave sports, AI will never leave writing or academia. Unless the data centers become too expensive to be sustained, the tool is here to stay. 

    The best thing I can do as a college writing instructor is craft assignments that defy AI use, such as field research, interviews, and personal reflections. I know full well that students will try to cheat the system and that I may have to have some blue book compositions to get a better handle on their writing, but we live in a high-performance world where often dangerous tools have to be harnessed. Otherwise, they will destroy us. 

  • Why I Need to Craft “Natty” Wring Prompts

    Why I Need to Craft “Natty” Wring Prompts

    About fifteen years ago, one of my students wrote a personal narrative about losing herself. She was in middle school. Her friends persuaded her to dress in extremely suggestive clothing with tons of makeup. In a revealing outfit that would have caused her mother to have a brain hemorrhage, she looked in the mirror and started crying. “I didn’t even recognize the girl staring back at me,” she wrote. “I wiped the makeup off, got back in my regular clothes, ran home, and cried in bed.” 

    I thought about that image as I started working on some writing prompts for the Fall Semester in which my eighteen-year-old college freshmen contrast writing with AI and writing “natty,” to use an expression that describes “natural” bodybuilders who don’t take performance-enhancing drugs. Generative AI can be a great writing tool, but it tends to overdo everything. Our writing becomes that crying overdone girl staring back at herself, or the bloated bodybuilder who is more cartoon than a celebration of the physique. In all of that generative AI, something gets lost. 

    Us. 

    We can hardly be blamed. Generative AI can feel like magic, taking a 300-word idea and blowing it up into a 2,000-word essay with vivid descriptions, highly stylized metaphors and witticisms, and Socratic rhetorical flourish. Eventually, we begin to believe we are the source of that 2,000-word essay. We flatter ourselves to our own detriment. We lose authenticity. We lose ourselves. We are that middle school girl staring into the mirror and not recognizing who’s looking back at us. 

    Using AI tools with my students for the last four years, I can say from personal experience that writing natty feels cleansing, like going off sugar-laden processed foods. Just as too much dessert causes diabetes of the body, too much generative AI causes diabetes of the mind and soul. We need to carve out natural writing spaces where we can connect with who we truly are before it’s too late. 

  • The Rise and Fall of the Influencer

    The Rise and Fall of the Influencer

    In her essay “The Extras Are Tired,” Ellen Cushing traces the astonishing rise of the influencer. Less than a decade ago, the word itself barely registered in the public vocabulary. The profession was still flapping uncertain wings. By 2023, those wings had become a full-fledged migration. Twenty-seven million Americans were earning money by producing online content, and universities were rushing to legitimize the phenomenon. Arizona State University now offers a bachelor’s degree in Content Creation, as though aspiring to become a walking selfie stick had matured into a respectable academic discipline.

    Yet the influencer economy already feels less inevitable than it did only a few years ago.

    One reason is technological. Increasingly, people seek advice from artificial intelligence rather than from smiling strangers who begin every video by announcing, “Hey, guys!” with the synthetic warmth of a flight attendant welcoming passengers aboard a delayed flight. AI may have its flaws, but at least it doesn’t interrupt your day to remind you to smash the like button.

    Another problem is more fundamental. Influencers possess a surprisingly short shelf life because their relationship with followers is built upon counterfeit intimacy. They manufacture the illusion that they are your trusted friend while simultaneously converting every friendship into a monetized sales funnel. Eventually, the performance begins to squeak. The smiles appear rehearsed. The vulnerability feels focus-group tested. The authenticity arrives with affiliate links.

    The public has noticed.

    Some clothing stores have become so weary of influencers transforming shopping aisles into film studios that they now post NO INFLUENCERS signs. The signs possess a wonderfully deadpan wit. They function almost like restaurant health inspections proudly declaring the premises free of rodents. Customers may now browse in peace, reassured that no one will suddenly begin recording a “Get Ready With Me” video between the denim racks.

    The signs are also an overdue act of cultural pushback.

    Influencers do not merely occupy public space; they annex it. They assume that coffee shops, sidewalks, airports, grocery stores, and boutiques exist as interchangeable backdrops for the ongoing production of their personal brand. Everyone else becomes unpaid production staff.

    This is a distinctly twenty-first-century form of vanity, one I call Protagonist Inflation: the progressive belief that one’s own life constitutes the central narrative while strangers exist primarily as background characters, audience members, or supporting cast.

    Cushing captures the phenomenon with characteristic precision:

    “Influencers today are a bit like plastic surgeons, or plumbers: On a societal level, many people sure seem to like what they do—they just don’t necessarily want to be right next to them while they’re doing it.”

    Exactly.

    Society may consume influencer content by the billions of views, but that does not mean it wishes to sit beside someone filming six takes of a cappuccino before allowing the coffee to become cold enough to qualify as archaeology.

    Much has been written about the jobs artificial intelligence will eliminate. We are warned that AI will replace writers, teachers, programmers, accountants, lawyers, and artists. Perhaps it will replace many of them.

    When it comes to influencers, however, I find it difficult to summon much grief.

    If artificial intelligence succeeds in removing one profession built upon convincing the rest of us that we are merely extras wandering through someone else’s movie, then AI will have performed at least one genuine public service.