Don't Lose Your Voice
Maintaining Humanity in an AI-Saturated World
Welcome to Polymathic Being, a place to explore counterintuitive insights across multiple domains. These essays explore common topics from different perspectives and disciplines to uncover unique insights and solutions.
Today’s topic addresses the challenges of an uncritical use of AI in writing. On the one hand, AI represents a transformative tool that can help unlock creativity, improve research, sharpen arguments, and challenge bias. On the other hand, it can lull the author into a sense of agreement, feeding their ego, and drowning their voice in the banality of the average. Let’s dig into the consequences and the solution.
Intro
I feel like I’m losing my mind. Over the last few years, I have kept coming across more and more articles that are all starting to sound the same. The same cadence, tempo, style, flow. More and more often, I get about 25% into an essay, and I see a convergence, it’s almost a resonance in my brain, a high-pitched whine where my eyes start to glaze as the markers of AI begin popping out of the text.1 As I scroll, the essay just keeps going without a clear focus, peppered with complex words that add nothing to the context. At that point, I close it down and move on.
My intuition tells me it’s all AI because I use AI for many things. One of my new favorite things is using Gemini on Android Auto to ask questions about history, geology, small town economics, random facts, and more on family road trips. It’s super insightful, like a personal tour guide, but after a while, you notice the pattern. If not managed, it just blathers on, reuses the same ‘dress-up’ words, and constantly telling me how my questions and its answers are amazing, unique, fascinating…
I know what generic AI sounds like because I train my AI to avoid it. And, to be clear, the issue isn’t with the use of AI, as I really enjoy using it as a valuable tool, and it can provide a Brain Boost if used properly. But those essays I’m reading? I’m not reading anything unique, which is fine for historical facts about Cloudcroft, New Mexico, and a basic overview of different bike computers, but not for true insights.
Because what’s really valuable is what you add to the writing. What’s your voice? Do you sound like you write? Are there fingerprints and personality that make it unique? Is there some quirk or counter-intuitive insight that you’ve discovered and want to share? I can confidently say that if there was, the uncritical use of AI buried it under the banal voice of an LLM. Worse, when it’s no longer your voice, it’s really no longer your ideas. You’re just agreeing with the output.
Losing Your Voice
A great example of this is when, in a recent article on Neurodivergency, Eva Redford responded to an accusation of it being purely AI-generated with:
“AI is part of my writing process the same way dictation, spellcheck, or text-to-speech are part of someone else’s. I’m autistic, and the cognitive load involved in translating complex thought into linear written language is very real. Tools that reduce that load are accessibility tools, not shortcuts. The ideas, framework, structure, and lived experience in this essay are mine.”2
Which is fair and a point I made in the past about Artists vs. AI, where I shared how AI can help unlock creativity:
Two examples from writing are an author I know whose husband has a fantastically creative mind but crippling dyslexia. He can’t write and struggles to work with writers to capture his ideas. Enter ChatGPT which has inordinate patience and can take the ideas and bring them to life so he can edit. The second example is an author I met who has a progressive neurological disorder threatening her ability to write. She uses AI to help capture her ideas while her body fails. It’s her imagination and creativity expressed through technology before it’s too late.
It’s not the accessibility that’s the problem. Those two examples are people who used tools to capture their ideas. They had conversations with the AI, they used voice-to-text, and had the AI pull threads together. They added their voice, and they worked with human editors to polish it up. What they didn’t do was rely on it completely.
This is a nuance that Eve Fairbanks nails in a recent Atlantic article titled The Biggest Tell That Something Was Written by AI:
“Now some authors tell me they’ve embraced AI as a “writing tool,” no different from spell-check or a laptop. The phrase is protean and euphemistic, covering everything from using ChatGPT to find a quote to having it compose a long essay based on a two-sentence prompt.”
The tell, she says, is when you “Look closely and you’ll see that every part of the text is not quite right.”
My visceral reaction described earlier, that high-pitched whine in my brain, is that every part of the text reads nearly perfectly, but is not quite right. There’s perfect punctuation, a lot of big words, but nothing there. That’s the common mode of an LLM and the way it formulates a general prompt. I have many hundreds of hours fighting with LLMs to eliminate this bias, and so I’m sensitive to it.
That’s what triggered my internal AI detector with Eva’s essay in question and many others including Chris Stephens, CPA’s essay here, and The Gifted Experience who attempted to rebut3 my post on True Neurodivergence with her’s titled When Someone Renames Your Experience For You4 all of whom admitted using AI to write them. What you’ll see is that there’s content, but it just doesn’t feel authentic. They’re all 5,000 words, blathering on and on, claiming to present a new view while offering generic blurbs of nothing but a marginal average.5
Part of what bugs me is how often the uncritical use of AI comes from the neurodivergent sphere. I’m harder on them because they claim to offer a different perspective, unlike the Algowhores or PSYOPs we’ve looked at before, where we aren’t told to expect more. They talk about their ‘lived experiences,’ or ‘gestalt language processing,’6 yet all start sounding convergent and typical. Worse, it’s not actually their lived experience; It’s an inanimate AI describing something for them, which was averaged from everyone else, and they thought it was close enough.
This happens because AI is incredibly synchophantic and actively strokes your biases like Wormtongue, replacing you with itself if you aren’t careful. AI also doesn’t Default to Truth; it will tell you things that you want to be true. Lastly, AI is trained on enough victim narratives to make it sound compelling to the ‘author’ and to their readers. The outcome is that all three of those ladies I mentioned proclaim a “neurodivergent” perspective, all claiming to be ‘gifted’ or ‘exceptional,’7 (Chris claims twice-exceptional), while they all sound like the same AI.
What’s even more fascinating is that Eva turned around, just days after my critique, and wrote: “How AI Systems Exclude Neurodivergent People by Design.” Which is suberbly ironic because that’s the argument I’m making against her use of AI. There’s no need for me to rebut those examples more from that perspective, so let’s look, instead, at the other consequence of losing your voice.
To truly understand her contradictions, consider that both essays are clearly AI-generated.8 She took the easy path, tossed in a couple of prompts, and pasted the result. She didn’t struggle to form, unform, and reform the logic underpinning Critical Thinking, and she ended up in logical contortions while also missing the valuable insights sitting in plain sight.
You may recognize that Eva gets close to understanding that AI is merely mathematical bias, on top of data bias, on top of organizational bias, but just misses that unique insight. The irony of her angst was already covered here, in Eliminating Bias in AI/ML, where, simply put, you cannot eliminate bias; you can merely bias the bias. Moreover, she misses that when you pathologize healthy personality, you tend to be excluded from normal engagement, which is another of her frustrations.
All of this occurred because, when you lose your voice to AI, you also risk ceding your thinking.9 Eva not only ended up with the banal average in her claimed experiences, but was also unaware that she was uncritically creating cognitive distortions. She’s pasting AI content instead of learning about the nuances that drive the challenges she seeks to rectify. She lost her voice, ceded her thinking, and lost the ability to find solutions.
AI is also very slippery. As Dean Bates recently describes in a recent essay, it’s not that AI is useless, but, to quote him, “No, the infuriating part is that it is useful enough to keep using.” This is the slippery slope with AI, especially when you aren’t looking at it with critical thinking. With its syncophancy, sophistication, and semi-accurate output, it can lull you into complacency and tempt you with the ‘easy button.’
The perverse outcome is articulated well by James Taylor Foreman who captured the risk in “We Know You’re Using AI,” where he shares:
“You burn bridges, in fact, when you indulge in not thinking. If you commit to this strategy, you’ll end up frenetically chasing your own tail, trying to spike your slowly declining popularity with people who, by definition, go away as they pay more attention to you.”
I’ll be blunt: when you lose your voice, you lose your thinking, and you lose the very thing you want people to follow you for: your humanity.
Finding Your Voice
That said, AI is a great augmentation tool, and even using it to compose a long essay isn’t necessarily bad. Jared Heymann hits the target with his struggle of balancing the tool with authenticity in the note embedded above. The tough reality is that there’s no right answer except for that struggle. The struggle is authentic.
Similarly, Dave Cline shared the note below showing that AI is the safe route, it’s not going to challenge what the guardrails deem politically correct, and it tilts toward justifying your inability to act instead of motivating agency to act. Both of these are anathema to creative writing or uncovering counterintuitive insights. Your voice should add something new to the conversation, not just mimic others.
For myself, I used AI to help me get past the blank page and start writing the first chapter of The Singularity Chronicles. The AI produced rubbish writing, but it got me writing and creating. As I went back to edit, I realized, about halfway through the book, I’d found my own voice, a cadence, a style. I then went back and re-edited the rest to pull that unique voice through. In the end, what the AI produced was completely eliminated by 120K words of my own, authentic voice. AI just provided the original grit to motivate the struggle that helped me find my voice.10
I’ve also found my own voice here on Substack. My AI-augmented spellcheck doesn’t like my style very much and recommends lots of edits. (but for some reason never flags that I can’t spell nuerodivergent) I ignore most of them, and what you read here is largely how I talk, and you are, quite literally, reading my brain flow… Even better, you’re seeing me wrangle to form, unform, and reform my thoughts. This last part is important because that’s not something AI does and, in my experience, can’t do. AI’s output is deceptively perfect but not quite right. My output is gritty and uncovers counterintuitive insights.
It’s also made me appreciate human writing more. Recently, I got a blurb from another author for a collaborative essay and loved it, but when they submitted a more polished version as we edited, it felt flat. They admitted using AI to touch it up, and I asked to keep the original because, even imperfect, I felt it had their voice, and I valued that more than perfection.
I’ve also reached out to fellow Substackers when I feel that what they published sounded like AI. If that was intentional, then OK, but it sounds like it. If it wasn’t AI… it still sounded like it.11 Some don’t care, and eventually I find myself tiring of their writing and falling away as James warned earlier. The others typically improve their writing as they find their voice and strive to highlight their unique contributions.
An added challenge, as J. Daniel Sawyer, who wrote The Pitch-Perfect Author: Voice Mastery for Writers, notes is that what used to be considered straightforward formal business writing style12 is now what AI emulates by default. That’s part of the attraction we discussed earlier, but when content creators pollute the commons with AI content, the resulting enshitification also drowns out professional writers who are writing to a formerly established standard, which forces us to find a new voice.
However, this is where the most counterintuitive insight emerges because you can use AI in a way few do: It can actually help find your voice, highlight your biases, and challenge you to think differently. For example, I use AI against my own writing to:
Produce a counterargument
Steelman another’s argument to make sure I’m understanding
Have AI challenge my biases or identify biases in my writing
Critique my voice, my tone, and my style
Highlight gaps in my logic
Simply put, you can use AI as your toughest critic and best coach at the same time. The key is that you have to critique the output and force yourself to introspect when it feels like something you want to hear because it loves stroking your ego and bias.
Other tips are, as J. Daniel Sawyer recommends: “First thing to do is to listen to how you speak and write already: read your writing out loud, record yourself speaking extemporaneously, record and listen back to your stories or essays. You develop an ear for what “sounds like you.” You learn your own verbal/writing ticks, your sense of idiom and turn of phrase, etc.”
This is what I do with these essays, and each one has a recorded voice-over read by myself to ensure just this. I’ve found my voice. I think it’s important that others find their own, and balancing between the extremes of pro or anti-AI13 is key to that.
Summary
So, I feel like I’m standing between two worlds. I embrace AI as a tool, and it has dramatically improved many things in my own writing, and yet, the more AI becomes ubiquitous, the more I use AI, the more value I’m seeing in the raw, but human writing that remains. What I’m avoiding is slop, whether human or AI-generated.14 What I’m trying to produce are human insights into who we are.
So, for those authors I mentioned and everyone else reading, the key is to embrace the struggle of writing. You have to accept the grit of imperfection; you have to find the line between Augmenting your Intelligence with AI and replacing your voice and thoughts with AI. I want to hear your voice. This is how we maintain our humanity in an AI-saturated world.
If you see someone who you think needs this message, please share this article. The more people we reach, the more we can rebalance our humanity. At a minimum please hit the ❤️ button above or below. This will help more people discover Substacks like this one where we explore what it means to be human.
Polymathic Being is a reader-supported publication. Becoming a paid member keeps these essays open for everyone. Hurry and grab 20% off an annual subscription. That’s $24 a year or $2 a month. It’s just 50¢ an essay and makes a big difference.
Further Reading from Authors I Appreciate
I highly recommend the following Substacks for their great content and complementary explorations of topics that Polymathic Being shares.
Goatfury Writes All-around great daily essays
Cyborgs Writing Highly useful insights into using AI for writing
Educating AI Integrating AI into education
Mostly Harmless Ideas Computer Science for Everyone
And yes, it does in include the em-dash, the “its not x, it’s y,” and more. No, it’s not exclusive to that. It’s where the point never really pops out and the essay never really goes anywhere. For a deeper analysis of the increase, check out Karen Spinner’s article I analyzed 16,000 articles to find AI writing on Substack.
As Kit Perez pointed out, “You cannot have a disability that requires accommodation AND call it a superpower.” That, plus the other contradictions, point to something much more complex in the simple story of neurodivergency that Kit and I explored.
It was actually such a poor article, with so many logical fallacies and AI contradictions, that Kit Perez used it as a great example of what to avoid with Identity Protective Systems.
The irony is so thick with this one where a human-generated essay offering counterintuitive insights is rebutted by an AI with the implication that this provides a better description of her lived experience.
It was too late to add this insight from Chris K. N. who recently posted this note:
This is a bastardized term from psychology that has been used in therapy for whole perspective analysis, but has been twisted to mean something different. What makes this one even more confusing is that it’s used as a claim of being ‘set apart’ and a super-power capability, yet it rarely offers anything to the conversation, especially when it’s just AI content. Often, it’s a claim for why they are misunderstood while becoming an excuse to avoid accountability to be understandable.
It’s interesting to note that gifted used to mean super smart, but now it’s used to describe pathologies we used to call retarded.
I should also point out that she’s dropped 5,000-word essays nearly every day, week over week, because that’s a major tell for AI in itself. In comparison, my longest essays don’t even crack 3,000 words, and I post just once a week.
Another layer here is that the claim of neurodiversity and access to that community typically eschews the tools of critical thinking that would help them all.
You can watch a short video of those insights below, or on YouTube. I’m not afraid to admit I used AI in the creative process because the book is an exploration of AI and what it means to be human. Can’t get more human-centric than that. Especially when it helped me find my voice.
After I scheduled this, I saw an author I appreciate, Holly MathNerd post this note on the topic. I get where she’s coming from and, as you can see in the further conversation, it highlights the quandry that Mr. Sawyer pointed out that LLMs are trained on Business Language. Which I trained to use myself. I just think it’s useful to not sound the same as the writers we poked at earlier where they enter a two sentence prompt and paste the output. Call me prideful, but there is something enjoyable about a unique voice.
I should add that this style was intentionally average, neutral, and objective. This is helpful for business documents, news reports, and more, but it intentionally mutes any unique voice. That’s why editorials in newspapers were popular because they had a different voice.
To be sure, the anti-AI crowd needs to find their authentic voice just as much as those we poked at today! Too often, they lack the same critical thinking skills in understanding how AI works, can be harnessed, and can be helpful that the pro-AI people ignore. Simply put, it’s a balance.
What’s funny here is that I started this essay based on a post I put on LinkedIn highlighting this challenge. The comments are something else and highlight the true ignorance many people have in what it actually means to be human! The sheer number of logical fallacies proves just how human they were and exactly why they’re threatened by AI. What amazes me is how they took a call to ensure we don’t lose our voice and accused me of being pro-AI and anti-artist. Did you know I wrote a book about how that creates problems? Check out The Singularity Chronicles!