Writing Isn't A Process. It's Many.
Let's be honest, the rise of AI in writing can feel a bit… unsettling. If a machine can churn out text with a few prompts, what happens to human creativity? What about the critical thinking and deep learning that the traditional act of writing has always helped us develop? We wonder if we're outsourcing the very struggle that leads to insight.
These are fair questions, and they touch on something important. If we continue to see writing as one single, monolithic, sacred act, then yes, AI will feel like an intruder.
But here’s a crucial insight, borrowed from cognitive psychology, particularly the work of researchers like Ronald T. Kellogg: Writing isn’t a process. It’s many. It’s a constellation of different cognitive tasks. Some of these tasks are profoundly developmental, directly benefiting our thinking when we engage with them deeply. Other parts? Well, they're more mechanical.
The challenge—and the big opportunity—lies in telling these components apart. Because when used appropriately, AI can be a powerful partner, especially for those parts of writing that are cognitively burdensome but don't always deepen our understanding. By strategically offloading some tasks, we can free ourselves to focus more intensely on the parts that cultivate critical thinking and learning.
A Quick Flashback: We've Always Offloaded Cognitive Work
It’s helpful to remember this isn't the first time we've adopted technology to lighten our mental load. Think about it:
The invention of writing systems themselves was a monumental cognitive offload. Suddenly, we could store complex information outside our own heads.
The printing press then took this externalized knowledge and made it widely accessible, dramatically cutting down the "search costs" of finding information.
Calculators, from ancient abacuses to the ones on our phones, took the strain out of complex arithmetic.
Even everyday calendars and checklists are cognitive aids. They free up our working memory from the burden of remembering every single appointment and task.
Each of these innovations allowed us to redirect our mental energy, creating capacity for different, often more complex, kinds of thought. Generative AI is simply the new frontier in this ongoing story, particularly for the intricate work of writing.
Understanding the "Weightlifting" of Writing: Cognitive Load
To see how AI can assist us best, it’s useful to borrow a concept from educational psychology called Cognitive Load Theory, pioneered by John Sweller. Essentially, our working memory—the mental space where we actively process information—is limited. Sweller breaks down the "load" into three types:
Intrinsic Load: This is the inherent difficulty of the material itself. Some ideas are just more complex than others.
Extraneous Load: This is the effort imposed by how information is presented or how a task is designed. Think poorly organized instructions or a distracting environment. This is the "bad" kind of load that diverts resources from actual understanding.
Germane Load: This is the productive, "good" kind of effort – the deep processing dedicated to actual learning, understanding, and making meaningful connections. This is the weightlifting gym of the mind where we build our cognitive muscles.
When we apply this to writing, we see a clear path. Thoughtfully used AI can slash extraneous load (think basic formatting, correcting typos, or even finding a commonly used phrase). It can also assist in managing intrinsic load (perhaps by helping break down a complex research topic into digestible sub-questions for initial exploration).
What’s the result? More of our precious human cognitive resources are freed up for germane load—the strategic thinking, critical analysis, creative synthesis, and deep reflection that are the hallmarks of valuable writing.
The Real "Workout": What Happens When We Write
Ronald T. Kellogg, in "The Psychology of Writing," details the demanding mental exercises involved in writing. He identifies a few core processes:
Collecting: Gathering information, from memory or external sources.
Planning: Generating ideas, setting goals, and organizing content. This is the deep "thinking on a subject."
Translating: Converting abstract thoughts and plans into linear text—the actual crafting of sentences.
Reviewing: Evaluating and editing at multiple levels, from mechanics to overall coherence.
Kellogg’s work underscores that the planning and reviewing stages are often where our most significant "cognitive effort"—our germane load—is and should be expended. This is especially true for many forms of writing, such as crafting non-fiction arguments, essays, or reports, particularly in the early drafting stages. In these contexts, we're often more concerned with the validity and novelty of our ideas and arguments, rather than whether we get each individual word perfect from the outset. It's in these phases, not just in stringing words together, that we forge new connections, clarify our understanding, and genuinely transform our knowledge. We often figure out what we really think as we articulate, structure, and refine our ideas on the page.
Of course, this emphasis can vary. For a poet meticulously shaping a verse, or a novelist deeply immersed in the texture and rhythm of their language, the "translating" phase—that careful, word-by-word selection—is itself a primary site of discovery and germane load. In such creative endeavors, the act of finding the exact right word is integral to the development of the idea itself. The key, then, is to understand where your most crucial cognitive work lies for the specific type of writing you're undertaking.
So, how can we use AI to supercharge these crucial human-led parts, especially where idea development is paramount?
Quick Strategy: The AI-Accelerated Writing-Thinking Loop
One of the big hurdles in traditional writing is the slow, often taxing cycle of getting thoughts onto paper. We're trying to compose, clarify, edit, and format, sometimes all at once! This can eat up our mental bandwidth, leaving less for the deep thinking involved in planning and reviewing.
Generative AI offers a way to reshape this, creating a faster feedback loop for your ideas:
You Articulate (Human-Led Planning & Collecting): Focus your effort on structuring your core idea; this is the essence of Kellogg's planning phase. Create a detailed outline. Record a voice note explaining your main points, drawing from knowledge you have already collected.
AI Generates: The AI takes your human-structured input and rapidly generates a coherent first draft. This offloads much of the extraneous load of initial sentence construction (part of Kellogg's translating phase) and can even help scaffold some of the intrinsic load of getting complex thoughts into initial prose.
You Reflect & Refine (Human-Led Reviewing at the Idea Level): Here’s the crucial shift. View this AI-generated draft as an immediate "mirror" reflecting your initial thinking, not as a near-final piece. Your engagement now focuses on higher-order concerns, a core part of Kellogg's reviewing stage, but aimed at the ideas themselves:
Are the arguments clear and compelling?
Are there assumptions to challenge or expand upon?
How does this framing compare to other ways of presenting the idea?
What new connections or insights does this draft spark?
This AI-accelerated loop allows you to "see what you think" more quickly and engage in more cycles of idea refinement. You’re spending your best energy on shaping meaning and developing arguments (Kellogg's planning and reviewing at a conceptual level), not just on the initial transcription (mechanical translating).
To see this process in action, let’s imagine Sarah, our researcher keen to share a complex new study on intermittent fasting and cognitive function through an online article.
Seeing Your Thoughts Faster: Sarah's Journey with the AI-Accelerated Writing-Thinking Loop
Imagine Sarah, our researcher, who wants to write an article explaining a complex new study on the benefits of intermittent fasting for cognitive function. She has a lot of information, but getting it organized and clearly articulated feels like wrestling an octopus.
Traditionally, she might stare at a blank page, try to write a perfect opening, get bogged down in sentence structure, and after an hour, only have a couple of clunky paragraphs. Her mental energy is drained, and she hasn't even gotten to the core of her argument yet.
Now, let's see how Sarah uses the AI-Accelerated Writing-Thinking Loop:
Loop 1: Getting the Core Down (Human-Led Articulation & AI Generation)
Sarah Articulates (Planning & Collecting): Instead of trying to immediately translate her thoughts into prose, Sarah opens a document and types out a rough, bullet-point outline – a core planning activity:
Intro: Why cognitive function matters, brief mention of fasting.
Problem: Current understanding of brain health often misses metabolic factors.
Study Overview: Describe the new intermittent fasting study – key methods, participant group.
Finding 1: Improved memory scores (cite specific data point).
Finding 2: Increased brain-derived neurotrophic factor (BDNF) levels (explain BDNF simply).
Finding 3: Reduced oxidative stress markers.
Mechanism: How might fasting do this? (Ketones, cellular repair).
Caveats: Not for everyone, more research needed.
Conclusion: Promising, practical tips for exploring.
She also adds a quick voice note to herself: "Really emphasize the 'cellular repair' part – autophagy. That's a cool mechanism people might not know." This is her human insight guiding the initial planning.AI Generates (Assisted Translating): Sarah feeds this outline and her voice note insight into a generative AI tool with a prompt like: "Write a 1000-word article based on the following outline and key insight, aimed at a general audience interested in health and wellness. Explain scientific terms simply."
Within minutes, the AI produces a coherent, if somewhat generic, first draft. It has sections for each of her points and even attempts to explain BDNF. The AI has handled the initial, often burdensome, part of the translating phase of writing.Sarah Reflects & Refines (Idea-Level Reviewing): She reads the AI draft, not as a finished piece, but as a mirror to her initial thoughts. This is a crucial reviewing step, focused on the ideas:
"Okay, the AI explained BDNF, but it's a bit dry. And it completely missed the punch on autophagy from my voice note." (Gap in translating her intent).
"The transition between Finding 2 and Finding 3 is weak." (Problem with the planned structure's flow).
"It listed the caveats, but didn't really explore why they are important or who specifically should be cautious."(Superficial planning/translating by AI).
"The overall tone is a bit too academic. I want it more encouraging." (Issue with the AI's translation style).
Loop 2: Deepening the "Why" and Improving Flow (Human-Led Refinement & Planning)
Sarah Articulates (New Instructions for Planning & Translating): She doesn't rewrite the whole AI draft. Instead, she gives the AI new, more focused instructions based on her review:
"Revise the section on BDNF to be more engaging, perhaps use an analogy. Expand significantly on the autophagy mechanism and its link to cellular repair, drawing a connection to cognitive benefits." (Guiding AI's translating and deepening the planned content).
"Create a stronger transition between the findings on BDNF and oxidative stress, perhaps by highlighting a common underlying pathway." (Refining the planned structure).
"For the 'Caveats' section, elaborate on specific groups who should consult a doctor before trying intermittent fasting and explain the potential risks more clearly." (Adding depth to planned content).
"Adjust the overall tone to be more encouraging and less like a research paper. Use more direct address to the reader." (Guiding AI's translation style).
AI Generates (Version 2 - Improved Translating): The AI processes these specific revisions and generates a new draft. This version is much closer to what Sarah envisioned for those sections. The autophagy explanation is there, the caveats are clearer, and the tone is better.
Sarah Reflects & Refines (Idea-Level Reviewing):
"Much better on autophagy! But now that I see it written out, I realize I haven't really connected the three key findings (memory, BDNF, oxidative stress) into a cohesive narrative. They feel like a list." (Identifying a weakness in the overall plan/structure).
"The introduction is still a bit bland. It doesn't really hook the reader." (The initial plan for the intro needs more work).
"I think I need a stronger call to action or a more practical takeaway in the conclusion." (Refining the planned outcome).
Loop 3: Weaving a Narrative and Sharpening the Takeaway (Human-Led Structuring & Planning)
Sarah Articulates (Higher-Level Restructuring - Advanced Planning):
"Re-draft the introduction to start with a relatable problem: 'Many of us experience brain fog or a decline in focus as we age. What if one solution lies not in a new pill, but in when we eat?'"
"Restructure the 'Findings' section. Instead of just listing them, frame it as a story: 'The study first revealed a surprising boost in memory. But how did this happen? Researchers then uncovered two key biological changes: a surge in the brain-boosting protein BDNF and a significant drop in markers of oxidative stress. Together, these paint a picture of a brain undergoing powerful rejuvenation...'" (This is significant re-planning for narrative flow).
"For the conclusion, add a small subsection: 'Simple Ways to Explore Fasting Safely' with 2-3 very basic, actionable tips, emphasizing consultation with a healthcare professional."
AI Generates (More Structured Translating): The AI incorporates these more structural and narrative changes.
Sarah Reflects & Refines (Final Idea-Level Review):
"Okay, this is starting to feel like my piece! The narrative flow is much stronger. The core argument – that intermittent fasting shows promise for cognitive health through specific biological mechanisms – is clear." (The plan is solid).
"The examples are good. The tone is right." (The translation reflects the plan).
"Now I can see a few places where I want to inject my own specific phrasing or a personal anecdote. The AI has given me a fantastic scaffold."
Sarah's "First Human Draft" Moment:
After these three (or it could be more, or fewer) AI-assisted loops, Sarah now has a solid draft that truly reflects her core argument and intended message. She hasn't spent hours agonizing over initial sentence construction (mechanical translating) or basic organization (initial planning). Instead, her mental energy has been focused on the high-germane load activities of:
Clarifying her own thinking (deep planning).
Structuring her argument effectively (advanced planning).
Identifying weaknesses in logic or narrative (critical reviewing).
Ensuring the message is engaging and relevant for her audience (holistic reviewing and planning).
Only now does she dive into the more traditional "human" work of detailed wordsmithing, adding her unique voice, personal touches, and ensuring every sentence sings. The AI didn't write her article; it helped her think through her article post much more efficiently, allowing her to arrive at a strong conceptual foundation much faster by streamlining the more burdensome aspects of initial translating and iterative planning.
This is the power of the AI-Accelerated Writing-Thinking Loop: it’s not about replacing human thought, but about providing a rapid, iterative mirror to refine it, freeing us to focus on the parts of writing that truly grow our understanding.
Now, What if Sarah Didn't Have This AI Partner? The Traditional Gauntlet.
Now, contrast this with the traditional writing gauntlet, a process many of us know all too well. Without an AI partner, Sarah would likely face that intimidating blank screen, wrestling with each sentence of her introduction as she simultaneously tries to recall information, frame her argument, and find the perfect words. This taxing mental juggling act would continue through the body of the post, where hours could be spent trying to synthesize complex research, organize content, and craft clear prose, often leading to a disjointed draft. The subsequent revision process would then become a heavy lift, demanding significant effort to untangle structural issues and clarify core ideas. Ultimately, the journey would be longer and more draining, offering fewer opportunities for deep, iterative reflection on the core message.
Crucially, with so much mental energy consumed by the mechanics of simply getting words on the page, the final piece might be less ambitious, as there's less cognitive bandwidth left for exploring more complex angles or pushing the boundaries of her thinking.
Leading the Human-AI Writing Partnership
By consciously directing AI to assist with tasks high in extraneous load (like initial formatting during the drafting phase), we conserve our own cognitive resources. This creates more mental capacity for the demanding, germane-load activities: insightful planning, critical synthesis during collecting and planning, crafting nuanced arguments in later-stage drafting, and rigorous reviewing.
This collaborative approach empowers us to use AI as a lever. It helps elevate our own thinking, deepen our learning, and produce writing that is not only created more efficiently but is also more thoughtful, original, and distinctly human.
The emerging skill for all of us is this discernment: knowing how to guide this powerful new partnership to amplify our own best thinking, ensuring we focus our energy where it truly counts.
