As a writer who has dedicated a career to the craft, I believe Artificial Intelligence (AI) is poised to fundamentally shake the foundations of the writing world. While this shift presents an existential challenge, it also creates significant room for skilled writers to succeed. It's not all "doom and gloom"—really good writers are going to be alright. The essential question is, how do we reconcile the disruptive power of AI with the enduring value of human writing? How should writers be thinking about AI? This article will explore how to integrate AI into your writing process, leverage it for idea generation, and ultimately, how to "AI-proof" your writing.

As a writer who has dedicated a career to the craft, I believe Artificial Intelligence (AI) is poised to fundamentally shake the foundations of the writing world. While this shift presents an existential challenge, it also creates significant room for skilled writers to succeed. It's not all "doom and gloom"—really good writers are going to be alright. The essential question is, how do we reconcile the disruptive power of AI with the enduring value of human writing? How should writers be thinking about AI? This article will explore how to integrate AI into your writing process, leverage it for idea generation, and ultimately, how to "AI-proof" your writing.

From Skeptic to Enthusiast: My AI Journey

My journey with AI has been one of rapidly diminishing skepticism. For six years, I ran a writing school, Write a Passage, teaching thousands of students how to build online audiences through consistent writing, publishing, and email newsletters, or by crafting deeply personal narratives. Back when we ended the school in November—less than six months ago—AI was less than 10% of the curriculum. Now, it would have to be the lynchpin of it, demonstrating just how quickly the landscape has changed.

My initial exposure to LLMs (Large Language Models) was years ago, in the "dinosaur age of AI," after reading a Stephen Johnson article in the New York Times about GPT-3. The lightbulb moment came when I began to understand how LLMs function at a technological level—predicting words and completing sentences.

Despite this early insight, I remained a skeptic. I remember debating a mentor, a multi-billion dollar company CEO, who was aggressively pushing his company to become "AI-first." I argued that current models hallucinated and weren't close to human capability. He urged me to "watch the rate of growth." This advice proved prescient.

My personal "AI awakening" truly began with a trip to Argentina, where I used GPT as a constant, instantaneous tour guide and researcher. By posing questions throughout the day—about art, statues, or local history—and reading the compiled reports at night, I was able to learn more effectively than with nearly all the human tour guides I hired.

The AI Writing Revolution: Speed, Competition, and Progress

The rate of AI advancement is staggering, driven by fierce competition. The models themselves are getting better and cheaper at a fast rate. My personal engagement with AI is up about 10x in the past year.

A key development has been the rapid emergence of high-quality models. In a stunning reversal of expectations from 2023, when GPT-4 seemed insurmountably ahead, 18 different companies released a model as good as or better than GPT-4 throughout 2024. This fierce competition is leading to falling prices and continuous innovation.

To ignore these advancements is, in my view, foolish. We are on the precipice of a new paradigm of writing. This isn't just theory; it's already impacting the industry:

• As of September 24th (presumably 2024), official stats showed that 18% of financial consumer complaints, 24% of press releases, 15% of job postings, and 14% of UN press releases exhibited signs of LLM writing.
• In my own life, a full half of what I read is now AI-generated, whether through conversations with LLMs or deep research reports.

AI as a Research and Learning Partner

The utility of AI extends far beyond simple text generation. Tools like OpenAI's Deep Research feature—which uses an advanced, unreleased "03" model—are generating tailored, multi-thousand-word reports based on highly specific prompts. For example, generating a personalized report on the flora and fauna of my specific Austin neighborhood during springtime took minutes and was more useful than generic Google searches or lengthy books. The writing itself, in these latest models, is even starting to show signs of voice and style.

The Future of Writing: How to Become AI-Proof

While AI is dramatically raising the bar for non-fiction writing, it is not an end to human authorship. The number of people who can gain an audience and make money by simply outperforming basic AI will fall considerably. The key to lasting success lies in differentiating your work.

The Two Pillars of AI-Proof Writing: Personal Connection and Unique Information

My heuristic for non-fiction writing that will last centers on two fundamental areas:

1. Personal Experience and Connection

The more a piece of writing comes from personal experience, the less likely it is to be overtaken by AI. This includes biographies, memoirs, and personal essays.

• Connection is the Antidote to Loneliness: A major reason we read is for connection—the human-to-human feeling of being understood. As David Foster Wallace noted, we read to "countenance loneliness."
• The Hollow Narrative: No one wants to read a personal narrative from a computer; it's a completely hollow, soulless enterprise. Stories that delve into your emotional journey, change of mind, and life arc—like the moving testimony of the co-founder of Wikipedia I recently read—are exactly the kind of writing that will endure.
• My Bolder Work: I am personally shifting to more personal and opinionated work, focusing on my story of becoming a Christian—a deep, complex narrative that AI cannot replicate.

2. Expertise and Data LLMs Don't Have

Writing that includes data, facts, or information that LLMs haven't been trained on will be invaluable. This falls under the "Two Es":

• Experience: Specific, lived experience provides context an LLM can't access, such as knowing the vibe of Austin, Texas, or the intricacies of running 200 live Zoom sessions for a writing school.
• Expertise: Cutting-edge, up-to-date information shared in tight-knit social circles, like current changes to the YouTube algorithm, often doesn't make it into the LLM's training data until much later.

If you have a lot of experience in a domain and write with expertise, you'll be fine.

The New Definition of Quality: Tailoring vs. Objective Merit

We need to re-evaluate what we mean by "quality" in writing. There are two components:

1. Objective Quality: The widely recognized merit of a piece of writing (e.g., the strong writing and research in The Power Broker).
2. Tailored Quality: How perfectly a piece of writing aligns with the reader's current interests.

Critics are correct that the objective writing quality of AI often isn't as high. However, AI excels in the second dimension: it's perfectly tailored to your interests at all times.

For example, a two-thousand-word Deep Research report about the March flora of my specific neighborhood is more useful to me than an entire Pulitzer-winning, generic book on the nature of Austin. The AI is shorter, more specific, and precisely tailored to my curiosity, making it a higher "tailored quality" read. This rise in tailored, AI-generated content is why I'm reading fewer things made by humans—time is finite.

The Enduring Skills: Taste and a Spiky Point of View

The core skills required for a writer to succeed, with or without AI, remain unchanged:

• Taste and Discernment: The ability to discern what's worth keeping and what's not. Whether you're generating words yourself or an AI is generating paragraphs, the vast majority is junk that needs to be cut. Michelangelo's advice for sculpting—to "remove everything that isn't what that final statue should be"—applies directly to writing with AI.
• A Spiky Point of View: A unique insight, an idiosyncratic belief about what is true about the world that very few people agree with. This is the Peter Thiel interview question applied to writing. AI is trained on consensus and will struggle to generate these bold, unconventional takes as well as a human. My friend's radical idea for a school that teaches everything from K-8 in two hours per day is an example of a deeply held conviction that forms the basis of compelling, AI-proof content.

Writing With AI, Not For AI

The difference between AI doing the writing for you and doing it with you is critical. No serious writer I respect believes AI can do the writing for them.

My Process: The AI as a Thinking Partner and Editor

When I write with AI, it acts as a co-pilot that enhances my human contribution:

1. Idea Generation via Voice: I speak out ideas during a walk and use a custom AI prompt to instantly turn the transcription into an outline or polished prose.
2. Instant Feedback Loop: I immediately ask the AI: "What are the weakest points of my argument? What stories need more information? What transitions were unclear?"
3. Character Development & Theory: I use the AI to teach me writing theory (e.g., what makes a good character from Hollywood and literature) and then interview me about my own characters to push me into generating detailed narrative.
4. Arguing with the Model: I use models like Grock's argumentative mode to intentionally challenge my high-conviction ideas. This instant, no-consequences pushback helps me clarify my thinking, find "treasure boxes of insight," and then summarize the key arguments and counter-arguments at the end.

The Future Will Not Be Like Chess, It Will Be Like Music

I believe AI's integration into writing will be like music sampling, not chess.

• In chess, people care about the human drama and rivalry, even though computers are superior.
• In music, we only care if the song is a "vibe" (the objective quality); we don't care if it was made with an acoustic guitar, Ableton, or samples. Sampling was once seen as "theft" and "not real music," but it's now completely normal.

Similarly, I predict that in 10 to 15 years, the only thing that will matter is the objective quality of the writing, not whether AI was used to help produce it. I'm not interested in the best writing that only humans can do—I'm interested in the best writing, period.

AI and the End of Slop

While many fear AI will usher in an age of slop (content where publishing consistently is more important than quality), I believe AI is actually the end of slop, by my definition.

• The Age of SEO Slop: The internet has been polluted by SEO-driven slop—like overly long cookie recipes with unnecessary backstories—because incentives were misaligned: Google rewarded time-on-page, encouraging creators to add "freaking nonsense."
• The Age of Consistency Slop: In the personal writing world I came from, distribution was king. Publishing a newsletter every single week, regardless of quality, was often more important than the content itself.

That age is gone. The competition from AI—which can instantly produce decent, tailored information—means the bar for human-written content is rising dramatically. The only way to rise above the AI is to produce something really good.

Leveraging AI: Memory, Context, and Customization

As a tool, AI offers profound capabilities in areas humans are weak:

• Memory and Context: Humans are terrible at memory. AI can easily remember conversations, meeting details, and context from years ago. Services like Granola AI can already record and summarize work meetings, allowing for instant post-meeting Q&A. This is leading to a major change in how I take notes: from notes for myself to notes for AI to read, prioritizing large, information-dense documents that exploit the LLMs' ever-growing context windows.
• The Consolidation of Power: AI’s ability to instantly scan and contextualize vast amounts of internal data (emails, memos, reports) will give leaders unprecedented oversight. The prediction that a CEO like Sundar Pichai could replace all 30,000 middle managers with "AI Sundar copies" speaks to a future where a company could operate as the articulation of a single, coherent vision.
• Diverging Models: The competition between companies like OpenAI (GPT), Anthropic (Claude), and Grock is leading to model differentiation. Grock focuses on up-to-date information and "unhinged" personality, while Claude might focus on coding, and GPT on qualitative writing. To stay on the cutting edge, you must pay for and use the latest models (e.g., Claude 3.7, Grock 3, GPT 4.5) to see where the technology is actually going.

The future is here, and it demands that writers adapt, not resist. By focusing on the unique assets of the human experience—taste, spiky insights, and emotional connection—and by using AI as a powerful thinking and research partner, writers can ensure not only their survival but their unparalleled success.