Key Takeaways
1. AI generates syntax, but only humans write
Generating syntax is not the same thing as writing.
Syntax vs. substance. Large language models like ChatGPT are probability engines that predict the next word in a sequence based on mathematical patterns. They do not possess a mental model of the world, nor do they have memories, intentions, or a commitment to truth. While they can unspool grammatically correct prose at lightning speed, they are fundamentally "bullshitters" operating without any concern for reality.
The intelligence illusion. We are prone to anthropomorphizing these machines because of their conversational interfaces and rapid outputs. However, calling this technology "artificial intelligence" is a marketing trick; it is more accurately described as automation. The machine does not "remember" or "review" text; it merely replicates patterns of language on which it has been trained.
Key distinctions:
- LLMs fetch tokens based on statistical weights, not lived experience.
- They are fundamentally incapable of distinguishing truth from falsehood.
- The "thinking" we perceive in AI is entirely projected by the human reader.
2. Writing is an embodied act of thinking and discovery
Writing involves both the expression and exploration of an idea, meaning that even as we’re trying to capture the idea on the page, the idea may change based on our attempts to capture it.
Thinking through writing. Writing is not merely the transcription of pre-formed thoughts; it is the very medium through which we discover what we think. As we struggle to find the right words, our ideas shift, deepen, and clarify in ways that linear thinking cannot replicate. Removing the struggle of drafting by outsourcing it to AI means abandoning the cognitive work of thinking itself.
The rhetorical situation. Every authentic act of writing occurs within a dynamic triangle of message, audience, and purpose. Writers must constantly make complex, non-linear choices to navigate this space, a task that requires human judgment and empathy. This process is not always easy, but the difficulty of writing is precisely what makes it satisfying and intellectually transformative.
The process of discovery:
- Ideas often emerge from subatomic "notions," "hunches," and "inklings."
- The struggle to articulate an idea is where genuine learning occurs.
- A finished piece of writing is the evidence of a completed cognitive journey.
3. Writing is feeling and a vehicle for empathy
Writing is also feeling, a way for us to be invested and involved not only in our own lives but the lives of others and the world around us.
Emotional resonance. Human writing is fueled by lived emotions—grief, joy, anger, and love—which are transmitted across time and space to the reader. When we write from a place of genuine feeling, we invite the reader into a shared space of mutual humanity. This emotional connection is something a chatbot, which has no body or soul, can never experience or authentically replicate.
The danger of boilerplate. When institutions outsource messages of condolence or congratulations to AI, they produce empty, clinical boilerplate that deadens our collective empathy. This "thoughts and prayers" style of automated text treats human tragedy as a transaction to be managed rather than felt. To resist this benumbing, we must respect the opportunity and responsibility of what it means to feel as we write.
The power of feeling:
- Lived experiences, like witnessing a loved one's final breaths, cannot be simulated.
- Writing about difficult emotions provides catharsis and deepens self-knowledge.
- Even mundane writing, like a grocery list or a student recommendation, carries emotional care.
4. Writing is a holistic practice, not a mechanical skill
A writer is a person for whom writing is more difficult than it is for other people.
The four dimensions. Like cooking or playing an instrument, writing is a practice composed of skills, knowledge, attitudes, and habits of mind. It cannot be mastered by simply memorizing rules or templates, nor is it a talent that some possess and others do not. When we write, we are practicing our practice, integrating these dimensions to solve unique communication problems.
Beyond the myths. Popular theories like the "10,000-Hour Rule" or "grit" fail to explain how we actually improve at our practices. True development requires purposeful, varied, and engaging practice that allows for play, experimentation, and the freedom to fail. We must reject the idea that writing should be easy or fast; the difficulty is what builds a robust practice.
Elements of a writing practice:
- Skills: The ability to conceive, draft, revise, and edit text.
- Knowledge: Understanding both the writing process and the subject matter.
- Attitudes: Curiosity, empathy, resilience, and a comfort with ambiguity.
- Habits of Mind: Engaging in both deductive and inductive reasoning.
5. Deep reading is the essential precursor to writing
Reading and writing are inextricable, and outsourcing our reading to AI is essentially a choice to give up on being human.
The reading brain. Reading is not an innate biological function; it is an adaptive behavior that must be actively trained and maintained. Deep reading fosters critical thinking, memory, and empathy by allowing our minds to join with the mind of another. When we read deeply, we are not just extracting information; we are engaging in a creative partnership with the author.
The threat of optimization. In a digital culture that values speed, we are losing our capacity for immersive, deep reading, replacing it with skimming and "bulletpointification." Treating books as mere data-intake vehicles for optimizing knowledge strips them of their intellectual and emotional value. If we outsource our reading to AI summaries, we choose to give up on the very experiences that make us human.
Processing vs. reading:
- AI can process and summarize text, but it cannot engage with it.
- Using AI to summarize books creates a form of "academic cosplay."
- Deep reading requires slowing down and allowing the text to trigger personal epiphanies.
6. The "engineering model" of education fails human learners
At the heart of education, at the heart of any learning, is engagement.
The myth of the robot tutor. Tech evangelists have long promised that "teaching machines" and AI tutors like Khanmigo will revolutionize education by personalizing learning. This engineering model breaks learning down into a linear map of discrete, testable skills, treating students like products to be optimized. However, this model has repeatedly failed because it ignores the social and emotional dimensions of learning.
Why machines fail. Students do not learn in a vacuum; they require trust, community, and a sense of belonging that only human teachers can provide. An algorithm cannot diagnose the underlying cognitive or emotional roots of a student's struggle, nor can it inspire them with genuine enthusiasm. "Personalized" software often results in deeply depersonalized, isolating experiences that contribute to feelings of alienation.
The limits of automation:
- Learning is non-linear, messy, and highly variable from person to person.
- Algorithms cannot replicate the nonverbal context clues that teachers use to adjust instruction.
- True education is a shared inquiry, not a march through a pre-mapped curriculum.
7. Standardization has turned student writing into "academic cosplay"
My students had been incentivized not to write but instead to produce writing-related simulations, formulaic responses for the purpose of passing standardized assessments.
The rise of bullshit. Decades of standardized testing and formulaic templates like the five-paragraph essay have divorced "schooling" from "learning." Students have been trained to write like algorithms, producing polished, coherent, but entirely soulless prose that lacks any trace of original thought. This "academic cosplay" meets all the official criteria for correctness but has no intellectual or educational value.
The cheating panic. Because schools have historically valued correctness and compliance over genuine thinking, students naturally view ChatGPT as a sensible shortcut. If the assignment itself is "bullshit" designed to check a box, outsourcing it to a machine is a logical response. Rather than seeing ChatGPT as a threat, we should view it as an opportunity to reconsider exactly what we value in student writing.
Renewing the classroom:
- Shift the focus from the final written product to the student's learning process.
- Design authentic writing experiences that require students to solve real rhetorical problems.
- Use reflection to make learning visible and hold students accountable for their own thinking.
8. Generative AI threatens to drown the internet in soulless "content"
One side effect from AI is that the corpus of human knowledge from mid-2023 on will have to be treated fundamentally differently than prior to 2023.
The content flood. Generative AI reduces the cost of text production to near zero, unleashing an avalanche of search-engine-optimized (SEO) garbage designed solely to capture clicks. This "content" is entirely divorced from communicative intent, existing only to generate advertising revenue. As a result, the shared human commons of the internet is being poisoned with misinformation and low-quality text.
The death of trust. From fake authors with AI-generated headshots to "obituary spam" that monetizes grief with fabricated details, the internet is becoming increasingly unlivable. When algorithms write for other algorithms to read, it becomes nearly impossible to find accurate, human-produced information. This disruption threatens the very economics of writing, making it harder for human writers to get paid for their labor.
The impact on writers:
- Traditional publishing models are collapsing under the weight of digital disruption.
- Scammers are using AI to steal the identities and reputations of established authors.
- Writers must find new, direct patronage models, like Substack or collectives, to sustain their work.
9. We must resist technological determinism and the "economic style of reasoning"
Efficiency is not a neutral measurement we necessarily associate with human well-being.
The efficiency trap. The "economic style of reasoning" dominates our discussions of technology, asserting that speed, optimization, and efficiency are the ultimate measures of progress. However, the most meaningful human activities—learning, writing, loving, and grieving—are inherently inefficient. We must resist the pressure to automate these activities in the name of productivity.
Technological determinism. We must reject the passive belief that because a technology can be built, its widespread adoption is inevitable. Humans have successfully banned and regulated technologies in the past, such as human cloning, when they threatened our core values. We are resolutely in charge of what we allow the "alien" intelligence of AI to do to our lives and our societies.
Reclaiming human agency:
- We must protect the dignity of human labor from being degraded by cheap machine substitutes.
- If a task can be fully automated by AI, we should question if it is worth doing at all.
- We must insist on public oversight and regulation of AI development to prevent systemic harms.
10. A framework for the future: Resist, Renew, and Explore
Rather than trying to predict the future of technology, this is a book about the things I know to be stable that ChatGPT cannot and will not change unless we lose sight of the unique human experience of writing.
A roadmap for action. To navigate a world with generative AI, we must adopt a three-part framework: Resist, Renew, and Explore. We must resist the encroachment of AI into inherently human domains, renew our commitment to our nature as embodied creatures, and explore the technology's genuine utility with caution. This approach allows us to maintain our agency rather than passively capitulating to technological change.
Cultivating taste and community. We must escape the algorithmic flattening of "Filterworld" by actively developing our own unique tastes and seeking out small, decentralized human communities. By valuing the spiky, weird, and messy nature of human expression, we keep the machines in their proper place. We must remember that our lives are experienced in the world of process, not outputs.
The ongoing journey:
- Treat the integration of AI in schools and workplaces as a public problem, not a private one.
- Find trusted human guides to help navigate the rapidly changing technological landscape.
- Remember that there is no terminal proficiency in writing; the struggle itself is the reward.
