You can disagree with everything a platform is doing and still have your subscribers, income, archive, and professional network sitting on it. Leaving is an option. Staying is also an option. But staying doesn't mean accepting every feature or treating a tech company's business strategy as a moral instruction for your creative life.

In this episode

Why we need to stop reacting to each new feature as if it came from a literary friend who knows us personally. Karen Hao's Empire of AI and the competitive pressure driving companies to push costs onto workers, artists, and anyone whose data helps build the systems. My theory, labeled clearly as a theory, that distinctively human content is becoming more valuable precisely as it becomes harder to identify. Six boundaries worth setting: your process, scanning, your archive, direct access to your audience, unpaid platform labor, and knowing in advance what would make you leave. Plus recent research showing that AI accusations track social gatekeeping more than they track actual AI use, which means writing badly on purpose will not save you.

HumanPrint homework

Write your platform terms. Pick the platform where you publish or build professional relationships and answer five questions: What does it genuinely give me? What do I give back in work, attention, data, and revenue? Which parts of my process stay private? What do I need to export or protect somewhere I control? And what specific change would make me leave? Then take one action this week. Download your archive, export your contacts, update your site, or write your private AI-use boundaries.

Spotting AI writing has become an internet sport. Someone sees an em dash or a tidy structure and starts a forensic investigation from the comments section, and now the platforms have handed everyone official tools to do it. This episode is about why "I can always tell" is almost never true, and why turning readers into unpaid AI investigators is a bad trade for everyone including the people who think they're winning.

In this episode

Why "seems like AI slop" is a process accusation, not a quality judgment, and what happens to actual criticism when we stop asking whether the argument held up. Three possible motives behind these features: money, engagement, and data. Substack's Pangram partnership sitting alongside its MCP connection that lets publication owners pipe their data into ChatGPT and Claude. Why human fingers on a keyboard are not an ethical standard. And the movement from statistical estimate to moral verdict, which is where the real damage happens.

HumanPrint homework

For one week, refuse to play AI detective. Don't scan anybody's work, don't accuse a writer, and don't use "sounds like AI" as your explanation for why something didn't land. Name the actual issue instead. Was the language generic? Was the argument unsupported? Was the story missing? Be specific. And notice what the platform is asking you to report, label, and classify on its behalf. You don't work for the detection department.

Substack just partnered with Pangram to let readers scan your posts and get a percentage estimate of how much was human-written. LinkedIn added a "Seems like AI slop" report button. So your options are now: submit to the scan, explain yourself in advance, or turn the scanner off and let strangers decide what that means. No thank you. Your creative process is yours before it's anybody else's information.

In this episode

Why a percentage assigned to finished text can't reconstruct the process that made it, and why even a detector with a very low error rate can't measure how much the writer actually thought. The phrase "outsourcing your thinking" and why nobody has agreed on where that line sits. Socrates in the Phaedrus, where an Egyptian king argues that writing itself will destroy memory and give people the appearance of wisdom without the real thing. Henry James dictating to a typist, Mark Twain's typewritten manuscript, George R. R. Martin still working in WordStar on a DOS machine. What twenty-five years of ghostwriting taught me about human assistance in writing, and why the presence of a human doesn't settle the ethical question either.

HumanPrint homework

Make a private process map for one piece of writing. Trace it from first idea to publication, including every tool, person, conversation, draft, AI interaction, and decision you can remember. Then mark three things: where you did your most important thinking, where a tool or person genuinely helped, and where you gave away a decision you should have made. Don't build this map to disclose it. Build it so nobody else can define your process for you.

WHY gives your writing fuel. HOW gives it a voice. WHAT is where most people lose the thread, chasing trends and algorithms instead of their own curiosity. My own interests - mapping human writing voices, the history of innovation, combinatorial creativity, scoliosis research - don't share one clean underlying question. I don't think they need to.

In this episode

The difference between stated preferences and revealed preferences, and why "my audience wants this" is a claim worth interrogating. My own scaling era, where I followed other people's formulas and ended up making work I didn't recognize. Why a book works like a science experiment - a question, an exploration, a conclusion that often just produces better questions. How my voice-intelligence system came out of decades of ghostwriting colliding with a master's program in measurement and psychometrics.

HumanPrint homework

Build a curiosity map. Blank page, every subject you return to voluntarily, nothing crossed off for being too small or strange. Draw the connections. Pick one that surprises you and write a page about it without touching Gen AI. Only then bring Gen AI in for structural supplementation and then decide for yourself what actually belongs.

Every time another "how to spot AI writing" post shows up, my response is the same: how do I know YOU wrote it? Before you accuse anyone else of using Gen AI, can you actually name what makes your own writing recognizably yours? Spotting Gen AI is the easy part. Spotting yourself is the harder, more useful one.

In this episode

The three places voice actually lives: rhythm, story, and humor. Staccato versus legato, and why removing every phrase you think sounds like AI doesn't automatically leave you with a self. What happened the first time I asked an early chatbot to write for me, and why I was the part that disappeared. The line between decorating a house you designed and moving into one you had no hand in building. Why reader rapport, even a little of it, belongs in almost every book, nonfiction included.

HumanPrint homework

Pick three pieces of writing from before you used Gen AI heavily, ones that still sound like you. Read them out loud. Make three columns: rhythm, story, humor. Then pick something you wrote recently and compare it. Don't go looking for proof Gen AI wrote it. Go looking for proof you did.

I started writing at seven and spent most of my childhood thinking of it as a talent, not a compulsion. Then a life-and-death medical situation sent me into nursing instead, and it wasn't until I came back to writing years later, through screenwriting, that I understood the difference. Talent can sit on a shelf. Identity follows you around whether you feed it or not.

In this episode

Why the writers who finish are the ones with a non-negotiable reason, and the ones who quit at the first believable exit usually never had real fuel to begin with. The difference between building the museum and building the gift shop, and why so many authors try to build the gift shop first. My own stretch of trying to write like a scaling business, complete with funnels and free downloads and marketing videos I couldn't watch back. Why your why doesn't have to be one sentence carved in stone, and can change by project. What structural supplementation means, and why Gen AI can help you organize your reason but can't hand you one.

HumanPrint homework

For one week, keep two running notes. In the first, write down every idea that gives you energy, curiosity, or a genuine pull to explore it. In the second, write down every task you think you should do… the ones you feel obligated toward. Then look at the language you used in each. That's the beginning of finding your fuel.

In May 2025, the Chicago Sun-Times ran a summer reading list with ten books that didn't exist — fabricated by AI, submitted by a freelancer who later said "I do use AI for background at times but always check out the material first. This time, I did not."

That sentence is the whole episode. Not a story about someone with no standards. A story about someone with good standards who, under deadline, skipped the one step that mattered.

In this episode

Why the people making AI mistakes usually aren't careless — they're tired, behind, and the output looked fine. The nursing background parallel: why procedures get drilled before the chaos hits, not during it. Why authors who wait for the cabin in the woods to write their book almost never finish it — and why the ones who integrate writing into a chaotic real life actually do.

Why the same principle applies to AI governance: the process you build when things are calm is the one that'll hold when life does what life does. Why faster and efficient are not the same thing when the speed is creating downstream problems you then have to spend time fixing, or explaining, or apologizing for.

HumanPrint homework

Make a two-column deadline list: left column is what AI can help you move faster with, right column is what AI cannot decide for you no matter how pressed you are. Fill it in when you're not under pressure — that's the version of you that knows what matters.

And if you're working on a book without a writing habit yet, that's the thing to build first. Make the boundary before the panic.

Each data point represents one pixel of the human experience. That's a stats idea, but it's also a writing idea — the word is the pixel, and the human experience that generated it is what gives it meaning. AI can produce words without the human experience underneath them, which means the pixels are there but the picture might not be.

In this episode

What a scatter plot and a book coaching session have in common — and why the most important thing in either one is sometimes the outlier, not the trend. What it felt like to read individual survey responses from fellow scoliosis survivors in her own research data, and why the individual dots told twenty stories the trend line couldn't.

Why when a book chapter feels "technically accurate but not actually mine," what's missing is almost always the same thing. The bulldozer questions that dig from message to root — and the moment when a client goes "oh damn, yeah, okay, now I get it" and the writing almost fixes itself.

The Mata v. Avianca case as the most vivid example of what happens when all the right pixels are there but the human experience that should be generating them isn't.

HumanPrint homework

Find a piece of your writing where something feels technically accurate but not quite yours. Ask the bulldozer questions: why does this matter to me, so what, why is this important enough to put in writing for strangers to read, do I have a personal story that actually connects to this idea? Dig until you hit the individual experience underneath the message. The trend without the scatter is just a line.

In 2024, Air Canada's chatbot gave a customer the wrong information about bereavement fares — and when he filed a claim, the airline tried to argue the chatbot was basically its own separate entity, not their responsibility. The tribunal said no. And that ruling is the whole episode: you don't get to outsource the judgment and keep the authority. Those two things travel together.

In this episode

Why most AI overuse starts with completely reasonable tradeoffs — "I'll fix it after," "just this once," "I'm sure it's fine" — and how those add up to a habit you didn't mean to form. The filmmaking adage "we'll fix it in post" and why it applies directly to what happens when you hand AI an underdeveloped prompt and hope for the best.

Two very different ways to ghostwrite a book — and which one most people are accidentally replicating with AI. The moment you stop developing ideas with AI and start approving ideas from AI, and why those are not the same job.

HumanPrint homework

Before you use AI for something that matters, write one sentence: "AI can help with ___, but I'm in charge of ___." Fill in both blanks. Then for one piece of work this week, mark every place AI helped and every place you made the final call. If the second list is thin, adjust the process. That's where your authority went.

This episode starts with scoliosis research and statistics, which may or may not be what you signed up for. But it's connected—because the same person who ran exploratory factor analysis in SPSS is also the person who had a titanium rod fused to her spine and somehow ended up doing acrobatics in her forties. Same person, different container. The problem starts when you confuse the container with the human being, and when you start applying professional restraint everywhere because you're trying to sound credible instead of being credible.

In this episode

How approval voice works like regression to the mean, as in, your sharpest, most specific sentences get revised toward the average until they could belong to anyone. The contradiction of wanting your writing to stand out while safely blending in. Why AI is very good at helping you sound acceptable, and why "make this more professional" is basically an invitation to move your voice toward the middle.

What data storytelling and writing have in common. The topic doesn't make the piece original, your relationship to the topic does. A 2025 study on AI-generated college admissions essays that found human-written essays contributed more new ideas, and the gap got wider as the sample grew. The difference between professional voice ("what does this context require so the reader can trust me?") and approval voice ("how do I avoid giving anyone a reason to question me?").

HumanPrint homework

Do an approval audit. Take something you wrote or something AI helped you revise, and find where the language got safer but less specific. Pick one of those sentences and ask: What am I actually trying to say? What have I seen, lived, studied, built, or learned that gives me the right to say it? What would I say to one smart person who already trusts me? Rewrite from there. Before you publish this week, ask yourself: Did I make this more me, or more normal?

People know when they're hearing Bob Dylan. The question is, do you know when you're hearing yourself? A lot of people have never really developed the habit of hearing their own writing voice, and AI makes it very easy to mistake fluency for improvement. This episode is about tuning into your own signal before you hand your writing over to be "improved."

In this episode

Why "finding your voice" is the wrong framing, meaning, your voice isn't MIA, you just haven't learned to hear it yet. What happens when AI gives you back something smoother and you feel nothing—and why that non-reaction is data. What twenty-plus years of book coaching taught me about the damage caused by trying to "sound like an author." Why voice develops through writing, not before it and the connection between HumanPrint and my first podcast, Your Daily Writing Habit. Same thread, different source of friction.

HumanPrint homework

Before you ask AI to draft or revise something that matters, record yourself talking through the idea for three minutes. Use the transcript as your starting point. Look for where you sound most like yourself—where your energy changes, where you get more specific, where you say something plainly that you would've overworked on the page. Write from that material. Then, if you bring in AI, tell it what to keep and what not to change.

AI has become the most rigid, literal editor in the world—determined to stamp out your unique writing messiness every time you color outside the literary lines. But some of your real voice lives in the roughness. This episode is about why the weird, unpolished parts of your writing might be the most valuable parts, and what happens when you let AI sand them off.

In this episode

Emily Dickinson's fight with her editor Thomas Wentworth Higginson over her "weird" punctuation and rhythm—and why AI is now playing the same role. Why the sentence that sounds "better" can be the wrong sentence if it no longer sounds like you. What I learned writing my own memoir, The Power of the Curve, about sharing the messy vulnerable stuff—and how my mess became a reader's miracle. The micro-decisions that happen in the friction between your rough draft and AI's polished output, and why those micro-decisions ARE your voice.

HumanPrint homework

Write something messy and YOU on purpose—the kind of messy that makes you nervous. Ask AI to polish it but NOT to make it into something that's not you. Share it with the world or just one person. See how it feels.

Three questions inspired by Emily Dickinson: What would a rigid editor "reject" about your natural writing voice? What's valuable about knowing that? And how can you push back?

In this episode of HumanPrint, Christine Whitmarsh looks at the line between useful support and outsourcing your own judgment. From core message to emotional framing, some parts of the process need to stay in your hands on purpose.

This week’s HumanPrint check: write down three things AI can help you do, then name the one thing it doesn’t get to decide.

In this episode of HumanPrint, Christine Whitmarsh looks at the difference between a strong voice and a polished one, and why the prettier sentence may not be the sentence that actually sounds like you. If your writing sounds smooth but strangely generic, that smoothness might be covering over the place where your actual point of view should be.

This week’s HumanPrint check: find one sentence that sounds impressive but doesn’t say much. Cut it, then write what you meant before you made it pretty.

In this first episode of HumanPrint, Christine Whitmarsh introduces the idea of drift: the subtle way generative AI can pull your writing, voice, and point of view away from your own human signal before you realize it’s happening.

This episode, HumanPrint itself, comes from a more useful place than the tired, black and white, pro-AI versus anti-AI debate. Our focus here is how to use the tool without letting speed, polish, and convenience genericize the parts of your voice that make your work recognizably yours.

This week’s HumanPrint check: before you publish, ask what sounds specifically like you, and what could have come from almost anyone.