The Importance of Voice: Or, How I Made AI Tell Cow Jokes in a Podcast
After interacting with a few personal bots based on real people, I started wondering: Could there be a GPT version of me? More importantly, should there even be a GPT me? And, what happens to GPT me when I inevitably abandon it, like I did this blog for years? Would it sit in some digital void, waiting for me to come back and ask it to help polish an essay or make sense of my rambling ideas?
The potential is certainly intriguing. What if this AI version of me could understand my own writing process? It could help clear the mental blocks, intuitively know when to cite research, or, better yet, when to throw in a bad cow joke (because, let’s be honest, I have a soft spot for those).
The idea feels oddly fitting since I just spent an entire year interviewing writing faculty for a research study on how they’re using AI to teach writing. We discussed invention, ideation, citation, voice, humor, and more. Which has made me ponder that for my own writing, I use AI tools for business writing, report synthesis, scripting multimedia content, Dungeons and Dragons quests and more, but I have not customized it yet to my own writing pursuits. So, here we go.
Horrifying and Hilarious GPT Name Suggestions
Let’s talk about the GPT name for a moment — because naming your digital doppelgänger matters. When I first started this little experiment, I asked ChatGPT for suggestions. Among the terrifyingly utilitarian choices were virtual assistant names such as “Thought Scraper” and “Brain Chisel.”
Horrified. I was horrified.
“Scraper” sounds like something that gets rid of paint chips or old gum under tables — not something you want anywhere near your delicate, messy, creative ideas. So I went back to the metaphorical drawing board. A squeegee felt like the perfect antidote — a tool that polishes, smooths, and clears without stripping away the layers that make writing meaningful.
Thus, Megan’s Thought Squeegee (GPT) was born
But Thought Squeegee? Why not?! It’s a reminder that creativity needs a tool that polishes ideas without scraping them bare — a kind of AI that enhances, not diminishes. My AI squeegee assistant will come back later, as it is going to help me in an experiment with spoken voice, written voice, and how humor plays into connection.
The Role of Voice in Writing: How Close Will AI Get?
For the last year or so, as I dove into the world of writing instruction and AI integration in higher education, one theme kept surfacing: voice. When we talk about “voice” in writing, it’s more than just the style or tone; it’s the essence of the writer — a fingerprint of thought, rhythm, and identity. Voice is what makes a piece of writing unmistakably yours.
During my research, this became a critical question. Faculty members repeatedly raised concerns about whether AI could support the development of a student’s authentic voice, or whether it would simply flatten creativity into generic, well-structured but soulless prose. (Please note: as I used AI to help write this blog, it just called itself soulless. Self-awareness? Maybe. Comic timing? Not quite.)
Now comes the real question: How close can AI get to imitating voice in writing? The answer, as of now, is: close enough to be quite useful, but not quite close enough to feel fully human. Even though I’m designing Megan’s Thought Squeegee to mirror my style and preferences, I know it will only capture a snapshot of my voice—a polished outline, but missing the quirks that make my writing distinctly me. The more I train it with quirks, jokes, poetic analysis, and theoretical frameworks common in my thought processes, though, the more it sounds like me.
Being the owner of several walled-off gardens for AI use (as I would heartily not support doing this within an open model), I uploaded some of my past research, creative writing, old blog posts, proposals, and critical citations. I also included works I regularly wanted to pay homage to, examples of tone and poetic anaylsis of texts, and — because we have come this far — the historical legacy of bovine humor.
With all this context, I asked my Thought Squeegee to analyze my writing style and guide me as I re-enter the world of public writing. What could AI tell me about myself that I hadn’t already considered? How could it help shape this experiment in blending human creativity with digital assistance? In true Inception AI fashion, here are a few AI-generated insights into my own writing process, written by my very own Thought Squeegee bot, posing as me:
“Humor as Discovery: I use humor not just to entertain but to unearth new ideas and connections. Humor becomes a tool for intellectual exploration, allowing me to delve into complex subjects with a light touch.
Interactive Voice: My voice is conversational and inviting, often using questions or personal anecdotes to engage the reader directly, as though we’re on a shared journey through the topic.
Experimentation and Flexibility: I’m not tied to one structure or style. I embrace a fluid approach, often blending humor, reflection, and technical analysis in a single piece. Thought Squeegee, in turn, had to learn how to mirror this elasticity.
Exploring the Meta: I frequently pull back the curtain, commenting on the writing process or the very technology I’m using, which adds a self-aware dimension to my work. It’s about staying reflective and continuously engaging with the tools and processes I employ.”
In the end, Thought Squeegee was more helpful than I anticipated. It confirmed what I knew but also offered a clearer lens into how I navigate humor, creativity, and intellectual curiosity. And as I continue refining it with more content and layers of my writing, I wonder how much more closely it will reflect my unique voice, especially when I move it outside of the written word.
Beyond Text to Literal AI Voice Applications
While the question of AI’s ability to capture written voice is a fascinating one, it’s not the only frontier AI is exploring. I’ve also been looking into AI voice and podcasting tools, like the new hotness Notebook LM, which takes the idea of spoken voice literally. These applications allow users to create podcasts or audio-based learning tools using AI-generated voices that sound, at times, eerily human.
The same dilemma applies here: Will AI-generated voices ever capture the full nuance of a speaker’s personality? At a certain point, will they all sound the same? Sure, they can mimic a tone or pace, but can they match the energy behind spontaneous humor or the subtle shifts in emotion that come with live storytelling?
Just like with writing, these AI tools can help smooth out the technical edges — they can make you sound more polished or handle repetitive tasks like drafting intros or outros. But they can’t feel the excitement in a conversation or pick up on the perfect moment for a sarcastic aside. Whether in text or voice, AI can assist, but it still falls short of creating something that feels undeniably, unapologetically human. But, it’s also getting shockingly good at this endeavor.
And, thus I present my GPT Thought Squeegee-assisted AI-generated podcast….
AI Patterns Meet Humor and How Cow Jokes Are Eternal
Speaking of humor — or AI’s attempt at it — here’s a sneak peek into how Thought Squeegee handled its first comedic venture. You be the judge: Was it moo-ving, or did it just land with a thud a bit like a tipped cow?
On the build side, I had to try this numerous times because AI wanted to only discuss the glories of AI, ignoring the entire massive history of cow jokes, cow joke examples, and cow jokes in popular culture content I uploaded as a separate notebook. To get the podcast AI to pick up a smidge of cow, Thought Squeegee helped me script out all the AI content embedded with cow content for it to pick up a hint, rather an odor, of cow in the final format.
Did the voices sell the humor? Did the AI miss out on some opportunities for puns and bad cow jokes to wrap it all up? (I yearned, yearned, for a final bad cow joke at the end.) AI is still learning the art of comedic timing.
At what point do these AI-generated voices start sounding like they’ve all come from the same canned GPS monotone factory? Sure, they’re polished, but they often lack the quirky imperfections that make humor land. But here’s the twist: for those looking to combine ideas, listen to conversations, and get a sense of a new topic area, AI like this could be a powerful learning tool — especially for reaching diverse audiences with different learning preferences. It may not be great at closing out with a bad cow joke yet, but there’s potential for something more, something unexpected, in how AI can synthesize larger ideas, humor, and specialized content for broader engagement.
Humans (and Cow Jokes) > Algorithms (For Now)
In the end, AI can help smooth the edges, but it’s the human touch that fills the gaps with meaning, voice, and — most importantly — humor. Whether it’s in writing or a podcast, voice is more than just coherent phrasing — it’s about presence, the feeling behind the words. Writing isn’t just about being clear; it’s about being heard (or…herd, ha). Podcasting, too, goes beyond simply stringing sentences together — it’s about letting your personality seep into every pause, every laugh, and every delightful tangent.
Creativity, at its core, thrives on imperfections, surprises, and human nuances — things AI, for now, can only mimic at surface level. Until AI can conjure a truly original joke or surprise us with the kind of creativity that only human experience can generate, we’re still in the driver’s seat. Creativity remains our domain, and as long as that’s true, AI will be our tool, not our replacement.
