I’ve always had this dream of being a writer. And not like a fleeting dream, something that occasionally pops up in the back of my mind here and there, but a meaningful, weighty dream. A thing I have thought about nearly daily for the last 10 years.
But here’s a problem: I have a significant case of blank page syndrome (it’s a real thing.).
When I sit down to write, I spend most of my time staring at the screen. Maybe I’ll type a thing before getting stuck on one word and deleting the whole bit. Maybe I won’t type anything at all and convince myself that there are other things I should be doing. Either way, I often end up with nothing except a visceral reminder that I am not destined to write.
And then LLMs showed up. These generative AI tools (think ChatGPT, Claude, and Google’s Gemini) have given me new ways to tackle a blank page. Specifically, these platforms have revolutionized how I stare at a screen write by getting me into edit mode much more quickly, allowing me to interact with text, identify what I don’t like, and play around with what I would say differently.
In this article, I’ll share the primary ways in which LLMs are helping me be a more prolific writer (by being a more prolific editor).
TTFD: Time to First Draft
A primary cause for my blank page problem is a deep desire for perfection. I not only want, I assume, that my first draft should be perfect. Groundbreaking. The next great American novel (or article, or LinkedIn post). I’ve struggled with this for as long as I can remember and, before I started using LLMs as a writing companion, I’d only encountered one meaningful solution: Brené Brown’s SFDs (shitty first drafts). SFDs, described in detail in Brown’s Dare to Lead, are drafts that are merely meant to get information onto a page so that the ideas can be played with, refined, improved, and occasionally scrapped. And LLMs are great at SFDs.
For example, when I sat down to write this piece, and before I ever opened a blank page, I opened Claude and provided this prompt:
Good morning, Claude! I want to write a brief article, 500-1000 words, about the power of LLMs (such as yourself) to help me in the writing process. Specifically, I want to talk about how it helps me get over some of my blank page syndrome and into edit mode faster. Could you please help me by generating a first draft of this work, including some key ways that LLM can help in the writing process, that I can then edit?
Mere moments later I had a first draft. Was it the next great American article? Not even a little bit. But it meant that, rather than figuring out what I wanted to say by pulling from a near endless void of words and ideas, I could see the concepts I’d like to discuss and the broad structure of a piece that I could work within. Less of a beach and more of a sandbox, I suppose.
An alternative here, rather than starting with a full first draft, is to work collaboratively with an LLM through the prewriting stage to see if it shakes anything loose. For example, my business partner and resident community AI expert, Jarad Johnson, often initiates his process by asking an LLM to generate an outline featuring key topics and relevant experts. In this case, my prompt might look something like:
Good morning, Claude! I want to write a brief article, 500-1000 words, about the power of LLMs (such as yourself) to help me in the writing process. Specifically, I want to talk about how it helps me get over some of my blank page syndrome and into edit mode faster. Before I begin writing, I would like to outline the key topics I intend to cover and, where possible, cite any experts or data that could enhance my position. Could you please generate an outline for me, with 3-5 key topic areas, featuring key authors/thinkers and sources?
In this approach, the LLM generates a structure in which I can create while also adding new touchstones for me to find additional information (e.g., authors and/or sources).
Writing Out Loud
One of the challenges to writing is the internal-ness of it: it’s all in my head and, when I write it, I often find myself reading the thing back to determine if the thing I wrote even makes sense (the importance of which is best demonstrated by this very sentence). And, more often than not, it does make sense because of course I wrote something that makes sense to me. But readers, including yourself, are not me, and the quicker I can get something out of my brain, onto a page, and run through another filter, the better.
LLMs can act as powerful collaborators in these moments. For example, I shared the above paragraph with Claude alongside the following prompt:
Hi again, Claude. I have written a paragraph that is substantively accurate but seems a little messy/wordy. Could you please review that paragraph and tighten it up for me, with a specific emphasis on making the wording more clear?
And here’s what Claude produced:
Writing can be challenging due to its introspective nature: ideas originate in the writer's mind, and upon putting them to paper, the writer often finds themselves re-reading their work to assess its coherence (a necessity exemplified by this very sentence). More often than not, the writing makes sense to the author, as they crafted it based on their own understanding. However, readers, including yourself, are not privy to the writer's thoughts. Therefore, the faster a writer can transfer ideas from their mind onto the page and subject them to an external filter, the better the outcome will be.
I am not likely to use that specific paragraph (in fact I didn’t use it), because it doesn’t sound like me. But, Claude’s version provided me with some helpful phrasing options, making my rambling topic sentence much more concise and finishing with a strong claim about outcome quality.
A separate but equally interesting path here is to use the LLM to generate counterarguments to my position. I have suggested that writing is difficult because so much of it happens inside that JELL-O™ sitting on top of your head, but it’s possible that there are arguments for the value of that hyper-internal process. I can test this (and really explore my position) by using the following prompt:
Hi there, Claude. I have a belief that writing is challenging because authors often only have their own internal monologues to use as editors during the writing process. Could you please help me understand if the opposite might also be true? That using your own internal filter is, in fact, good for writing? And, if that’s the case, what authors or experts take that position.
For what it’s worth: when I use that prompt, Claude provides me with a coherent explanation of the value of using one’s own internal monologue to guide the writing process (it creates a unique voice!), as well as a list of authors (e.g., Anne Lamott, Natalie Goldberg) and their works to learn more about how to harness my inner voice while writing. It then reminds me that a blend of my unique voice during the drafting phase and external reviewers during the editing phase is likely to produce the best result.
The Things We Don’t Think About
The last meaningful way that LLMs have helped me write more (and hopefully better) is by acting as a topic generator for the things that I might have missed. For example, after finishing the previous two sections of this article, I returned to Claude with this prompt:
Claude, these suggestions for how LLMs can help me overcome blank page syndrome are great. I would like to offer readers with more reasons why LLMs can be useful writing companions. Can you please generate a list of 3-4 additional reasons that LLMs are powerful writing companions, providing a coherent explanation for each?
Claude provided the following three suggestions (with descriptions shortened):
- Rapid Research and Fact-Checking: LLMs can assist writers in conducting quick research and fact-checking while writing.
- Enhancing Vocabulary and Style: LLMs can help writers expand their vocabulary and improve their writing style. By offering alternative words, phrases, or sentence structures, LLMs can help writers find more precise, engaging, or eloquent ways to express their ideas.
- Exploring Different Perspectives: LLMs can provide writers with different viewpoints and angles on a given topic, helping to broaden their understanding and consider alternative perspectives. By asking the LLM to generate arguments or opinions from various standpoints, writers can gain insights that they may not have considered otherwise.
Each of these functions expedites the process of content generation so that I, a world-class critic capable of telling anyone and everyone how much I don’t like something, can spend my time editing.
Large language models have revolutionized my writing process, acting as a powerful tool to overcome my blank page syndrome. By collaborating with LLMs, I can generate (shitty) first drafts quickly and dive into editing mode much faster. So, if you’re like me and you find yourself knowing that you could be writing but unsure of where (or how) to start, find an LLM you like and let something else start for you.