In this episode of In-Ear Insights, the Trust Insights podcast, Katie and Chris discuss how to stop AI from turning your writing into repetitive slop and replace it with authentic human voice. You will discover why AI drifts into repetitive phrasing and how to stop it. You will learn to measure your unique writing style with simple numbers that lock in your voice. You will apply a structured editing process that transforms machine drafts into polished content. You will gain confidence to command AI tools without wasting hours on endless revisions.
00:00 – Introduction
02:15 – The AI writing frustration
06:30 – Measuring your voice
12:45 – The five performance steps
18:20 – Building your writing blueprint
24:10 – Call to action
Take the new course at:
https://academy.trustinsights.ai/courses/ai-for-writers
Watch this episode to finally break free from generic AI output and start writing with total control.
Can’t see anything? Watch it on YouTube here.
Listen to the audio here:
- Need help with your company’s data and analytics? Let us know!
- Join our free Slack group for marketers interested in analytics!
[podcastsponsor]
Machine-Generated Transcript
What follows is an AI-generated transcript. The transcript may contain errors and is not a substitute for listening to the episode.
Christopher S. Penn: In this week’s In Ear Insights, let’s talk about AI for writing and for writers. We have seen no shortage of people talking about AI watermarking and all this stuff and how you can tell whether somebody’s using AI for writing or not. And we at Trust Insights have put together a new course, Trust Insights AI for Writers, for how to get AI to write better and not coincidentally, help you as a human also become a better writer. So, Katie, to start off, what are the things that when you are writing with the assistance of AI, what are the things that sort of you wish AI would do better?
Katie Robbert: You know, I wish it would listen better. And by that I mean I feel like you can craft a really strong prompt. You can say, here are my writing samples, here are things that I don’t want you to do. And it kind of just like freewheels and does its own thing anyway.
And I feel like that is frustrating for a lot of people. So, you know, I really try to write the first draft of things myself as the human and then know, and I’ve talked about this in the newsletter and on pod on our podcast where I bring in AI is to double check with like our ICPs or to use it as an editing tool. But then the editing gets carried away.
You know, I’m not an editor and grammar is. I would like to say that the public school system failed me. You know. So I’m like a half decent writer. I have good ideas and I write in a stream of consciousness. And so I need tools or human editors to help me clean things up.
And this is where I look to generative AI because the team doesn’t always have time to like fully edit my stuff. But then when I read back what the edits are or what the suggested edits are, I’m like, where did this come from? Or why did you make up a whole anecdote that never happened, but you’re saying it authoritatively?
So I feel like the hallucination is the big thing. And the, you know, depending on the large language model, each model has its quirks in terms of the way that it writes or the way that it, you know, quote unquote, articulates thoughts.
And so I think one of the things that you’ve shared about Claude, for example, is it likes its, you know, things in threes, like three punchy points. And it’s very much a tell that, like, oh, that’s a Claude thing. And I found that when I have, you know, an AI assistant help edit my stuff, it turns it into a lot of that, like the individual sentences, the three punchy points, like the this and this, then this or this or, you know, those kinds of ways that it writes.
And that’s not what I, the human had put in. And I’m just sort of sitting there exhausted, like. But I just, I need this edited and I need it coherent and is it good enough? But did it change too much of my own human writing? So I think that, you know, when I think about using AI for writing, that’s what I’m personally struggling with is, you know, I still want to be the one writing it, but, like, I need someone to help me polish it. And the polish is like subpar.
Christopher S. Penn: Why do you think AI does that? Why do you think AI behaves the way it does and turns original writing into something that sounds like slop?
Katie Robbert: Oh, gosh, if only there was a course that was going to tell me the answer to this question.
Christopher S. Penn: There’s something else that will also tell you the answer to that question. And that happens to be the fifth P.
Katie Robbert: I should have guessed that one. Sneaky, sneaky. The framework at Trust Insights is purpose, people, process, platform, performance. Chris, you’re specifically talking about performance. And I think that where a lot of us get caught up in prompting these large language models is we think we’re being clear on the performance, but we’re not as clear as we could be.
And that’s where the frustration sets in and that’s where we want to throw up our hands. And so the perfect purpose could be, I need you to edit this, you know, five thousand word essay. I need you to look for spelling and grammar and, you know, a cohesive thread like all those things. People, here’s my audience, here’s my authoritative voice, here’s my samples process. I want you to go through this and just list out the changes. Don’t change it for me, platform. This is going to be published on my blog, which is hosted here, and I’m going to add images in these places.
And then performance. We typically think of performance as did we get the polished thing from our purpose? But it sounds like we are missing a lot of opportunity in the performance part of the 5Ps to really spell out what we need.
Christopher S. Penn: And that is the premise of the new course. The biggest chunks of the new course that we have are twofold. One, we spend a lot of time on research because good research leads to better outcomes typically. And two, we spend a lot of time on math, which is your average writer is like, But I started with the premise for this course, that writing is code.
If I put nonsense words together, you’re like, did you just get hit in the head? Like, what happened? Did you actually put decaf in the coffee maker this morning? If I don’t say words in the right order in a statistically predictable pattern, you have no idea what’s going on. You might say, these tests, coverage, adding branches, empty. Like, what? What does that mean? That’s word salad.
Language follows patterns, and those patterns are predictable. And the reason why AI writes the way it does is because it’s choosing the most probable patterns, even when it doesn’t sound like you. So the first thing that we have to do is give AI performance, right? To say, this is what success looks like. And it has to be in a tangible form.
The percentage of passive voice that you use in a text when you write as a human, how much passive voice do you use? The number of sentences that begin with a noun or a pronoun. What percentage of your copy is that? Is that the number of EM dashes that you use naturally in your text as a human? What is that? Our friend Anne Hanley says, I use the EM dash because I’m an actual writer, but I don’t use it in every sentence.
And where AI typically goes off the rails is when it knows that a construction is probable, like using EM dashes, like using triadic rhythm, like using bicolon or isocolon. And it says, hey, I’m going to use the most probable things. But it has no concept of frequency, so it overuses it. And you get, it’s not this, it’s that in every single sentence or in, you know, Claude in particular loves bicolon. It’s. It sounds like a drum beat. One and two and one and two. And you’re like, could you please vary the beat?
Katie Robbert: Right?
Christopher S. Penn: When you look at a, like a slide deck Claude generates, everything is bicolon, all the headings, you know, this and this, sharp insight and this. And you’re like, my God, this is so mind numbing to read. If we give AI analysis of how our writing to begin with and say success looks like this set of numbers, now go right, then check your work and compare what you wrote versus what the blueprint is and it will go, oh, I didn’t do this at all. Like, I use 82% passive voice. Yeah, go back and fix it. But the fifth P in the 5P framework by Trust Insights is so important. We have to establish what success looks like for it so that it mathematically can go back and fix its code.
Katie Robbert: So let me ask you this question though, because I will give Claude like samples of my writing and say this is what it’s supposed to sound like. Am I doing it wrong? Because I have all of these samples from like literally years and I feel like Claude or a large language model, you know, is trying to evolve my writing so that my writing fits its format, not its editing, to my writing. Like, I feel like that’s where I’m struggling.
Christopher S. Penn: You are not doing anything wrong except you are asking it to count and it can’t count. And so even though it will analyze your writing as a language model, it has no clue of how to count. One of the things that’s in this course is probably worth the price of admission alone is a Python script that it has to run on a writing sample you provide that will do that fingerprint mathematically not letting the language model try to count, because language models can’t count.
The Python script goes through and it counts in your original sample, this is the percentage of passive voice that you use and it writes it down as a fingerprint, as a file in your language model of choice. It works in ChatGPT, it works in Copilot. It tested in all the systems. It can then rerun that script on its output and say, initial sample, 4% passive voice, my work, 18% passive voice revision loop. I need to go back and keep revising until I hit this number.
But it can’t count that by itself. It needs the support of actual code to do it. And that’s what’s in the course is pre baked. Nobody has to be coding, no coding involved. It’s bundled in. But you would drop that into your Copilot or your ChatGPT or your Claude and say, this is how you’re going to measure yourself. Yourself, you’re going to do the fingerprint and you’ll reuse that fingerprint and then you will go back and you will count using this script. And that’s, you know, again, you’re not doing anything wrong. It’s just the average non technical user doesn’t think, hey machine, I remembered you can’t count well.
Katie Robbert: And I say, okay, so that’s an interesting distinction because so up until now, you know, I’ve been saying like, hey, this is the sample of my writing. And you’re right, it absolutely, it takes what I give it and it like, is like, oh, you said to do this, let me make sure that I do this. And by making sure that I do this, I’m going to do it 16 times out of the 18 sentences, but I’m also going to do this in 15 of the 18 sentences. And so it’s taking everything it knows about my style of writing and trying to jam it all into one sentence. And I’m like, whoa.
Like, yes, grammatically it’s correct, but now it’s garbage because that is not at all what I wrote. And I sort of. I feel, it’s not that I feel like my hands are tied because I can’t fix it myself, but like, the reason I turned to a system for help is because I’m not an expert editor. And so I miss things that an editor would find. And I need that, like, for me, I need that kind of support of like, hey, you started a point over here and then you dropped it halfway through and you made a different point at the end. Like, you gotta pick a story and stick with it.
Christopher S. Penn: And this is where these tools, these AI tools simply by themselves cannot do that. Like, they just do not understand. How do I, how do I even count? So I’ll show you an example of one of the things that is bundled in the course. And again, the average user does not need to look at this. The average user is not going to. You will just drop the file and say, machine, off you go.
But it will look at thing. There’s functions in this. It says like, look at the rate per word of the this kind of word. For example, it’s often said that good writing relies on relatively few adverbs. Adverbs are words that end in ly, you know, actually, etc. AI loves adverbs, obviously, actually, and stuff like that. Which also sounds condescending.
Katie Robbert: Yeah.
Christopher S. Penn: And so if your writing style uses almost no adverbs, when you do a fingerprint, it will say, hey, your adverb rate per 1000 words is like 1%. And so when it goes back and counts, it’s revision, it’s edits that you had it make. And it goes, oh, I used 9% adverbs in my revision. But the target, the performance says 1%. I need to go back and fix my work. It’s a diagnostic. And that’s what’s missing from all of our prompts, because it can’t do that.
That’s what’s missing from every single AI for Writers course I’ve ever taken or every session I’ve ever sat in. Nobody thinks of writing as a system of measurement, of analytics. And therefore, when AI just follows its own internal process programming the probabilities that it generates, we’re all like, why is this keeps coming out like slop? Why can I not prompt this thing to sound like me? It’s because it can’t count.
And so the cornerstone really, of this entire course that we’ve created is let’s give AI the tools it needs to count, let’s give it the measures that it needs to count, and then let’s give it clear guidelines. You know, one of the things that is in the toolkit is a. Again, this is another piece of code the user, you. The user will not use. This is built into a skill that you just install, and there’s instructions on how to install it. But it will say, if you have this cadence, don’t do that. Here’s how you do it instead, right? If you do short, long, short, long cadence over six consecutive sentences, you’re writing us. You’re writing a dead drum beat. Dead. Don’t do that, do this instead. Things like that.
So, for example, if you look at AI writing ChatGPT, Claude, Gemini, and you just take a step back and you look at the page, all the paragraphs are about the same length. They’re all. They all kind of look like the same gray rectangles on a page. If you were to step back and stop looking at the letters and just look at the shape. If you look at your human writing, there’s a good chance that it’s. There’s some paragraphs are real short. Maybe it’s even one word, like an emphasis one like, no, don’t do this. Right? And that’s just its own paragraph.
That frequency of change in paragraph length is something that you can measure. Machines don’t know to look for that. Machines don’t even think about that. And so if we give them the tools, the counting tools to go. Katie tends to alternate her paragraph length. Sometimes her paragraph length is 11 sentences, other times it’s one. I should replicate that general pattern. Because once you give AI a pattern, it’s like, oh, I know how to do patterns. I can do this. And it goes off and it creates it.
Katie Robbert: I mean, I have a lot of thoughts and comments, you know, and so like, the big elephant in the room is, you know, why are we teaching people how to train the models to write when there’s real writers out there? So, I mean, that’s a big question. So I’m just going to like, put that, like, stick it up here for a second. When I think back to like high school and middle school, for example, we generation were taught the five paragraph writing. But for a lot of us, this is like, this is how were taught to write.
So the first paragraph is your opening argument. The second, third and fourth paragraph are your supporting reasons for your argument. And the fifth paragraph is your conclusion. And so a lot of us who were, you know, that was like drilled into our heads on our like yellow piece of paper with the green lines, like that’s how you’re supposed to write. And so I think what’s often frustrating for someone who writes their own stuff is because were taught to structure our writing in a certain way. It can come across as well. AI must have written that because of the structure. It’s like, no.
My sixth grade middle school English teacher, when slapping rulers on tables was legal, scared the bejesus out of me and told me, this is how you have to write. And so I write the way that AI writes because it was drilled into my brain. This is how you write. And I guess I’m wondering, so when you’re saying like pattern recognition, like it learned these patterns from us. We taught it the patterns it did.
Christopher S. Penn: But it is averaged together everything. And that’s why it often comes out so different than the way an individual writes. Because everyone has their own pattern distortions. Everyone has words they like, everyone has words they don’t like. Everyone has words, life experiences that will show up in your writing. AI is averaged all of that together into. And then what it does is it spits out the highest likely probability except for when it’s using watermarking. And so the five paragraph essay and that kind of blocky set. Yeah, it’s going to do that because the majority of writing it has seen on a bell curve is exactly that.
If you look at the, you know, an earnings report or a press release, it is exactly that dead metronome of boring writing. The thing about writing to that. And we say this in the beginning of the course and we’ve said this in many different places. Good creative work that’s interesting is low probability. Right. You, the way you write should be surprising and different than what the way that middle school teacher taught you to write. Right? I will. There’s all sorts of expressions I’ve used for this.
But if you say, you know, this is a, this works like a Prius and other people the average is this works okay. Right. This works okay. It’s boring. This works like a Prius, has a very specific connotation and A mindset behind it. And so what we want AI to recognize with the tooling in our course is recognize how the. In our individual style looks and replicate that specific pattern, not the general patterns. You’ve been trained on the general patterns that you will generate without these very rigid mathematical guardrails.
Katie Robbert: Okay. One of the things that I’ve noticed and I’ve seen, and actually, this is a phrase that you use a lot. And so I guess I’m sort of asking, like, have you adopted this from AI or is it a phrase that people use and AI has adopted it is something that I see a lot in my conversations with a large language model is. And here’s the shape of the thing, and here’s the shape of it, and here’s the shape of the problem, and here’s the shape of the challenge. Well, that’s not the same shape as it was before. And I’m like, why are we talking about shapes? And to be fair, Chris, you say that a lot, but you are not in my instance of a large language model. And so I guess my question is, have you. No, you’re not. Have you brought that from working with large language models? Because that’s not a phrase that you used to say before. But I find that, like, there’s these little ticks and, you know, quirks that these large language models have, and each one has a different one of ways that it phrases things. So you’re talking about, like, the 1, 2, 3, the cadence, but then there’s also these descriptors, and I find that really interesting. And I feel like if a human is not paying attention, that very easily slips into a lot of your work.
Christopher S. Penn: It. And it slips into how you think, too. Like, I find myself when I’m writing now, I go, oh, that is negative parallelism. And I will even call that out. Like, in a LinkedIn post I published this morning, I say, and to quote one of Claude’s favorite constructions, it’s not this, it’s that. Because negative parallelism, which is a form of bicolon, is. Is something that the language models use. But writing is a lot like. It’s a lot like nutrition. You become what you eat.
So if you are reading AI generated text all the time and. And you start seeing that’s a sharp insight, that’s a sharp angle. It’s going to influence you. And so, yeah, I probably have picked up things because I’ve spent so much time wrestling with these tools and trying to diagnose how they do their constructions that, yeah, some of it is going to influence how I write and speak and even think. And one of the things that linguists, if you read this one, the discourse online, the, one of the things that linguists are very concerned about with AI generally is that it’s sort of a flattening of language, that the language, the English language itself is morphing because of the influence of AI on it.
When you, when I saw, I think the stat was two out of three pages on the Internet are now written solely by AI. That is going to have a, a shaping effect on how we read and how. And when you see, you know, I think it was something 10 out of 1, 9 out of 10 new books submitted on Amazon was written solely by AI. That’s going to change your language. That’s going to change how you read and how you write is how you think.
Katie Robbert: Yeah, it’s interesting. So let’s say I’m a brand new learner. I want to sign up for the AI writers course. What are some of the big things that I’m going to learn? I believe you’re introducing a new framework which is called craft. Can you speak a little bit about what the craft framework does and if I’m someone who’s trying to help my writing, what that means?
Christopher S. Penn: Well, okay, so the CRAFT is a subset of the 5P framework, right? So the whole course is Craft fits in process. It’s a breakdown of process. But fundamentally the whole course itself is actually structured on the 5P framework. I’m the major lessons are literally right out of the 5P framework by Trust Insights because it’s the best framework for pretty much everything. CRAFT is a subset of process, which stands for create. You need to come up with the idea because AI is not going to do a good job.
Then you need to do a buttload of research to inform the idea. Then you need to architect the work itself and we’ll talk about process decomposition, which is which we borrow from software development. I basically took the learnings of some software development and turned it into a writing course. Then you have what’s called fabulate, which is basically a machine. Here’s how to go and do this. How you build the. The plan for a tool to go do it. And then the last part, which is the hardest part, is the tuning part where the machine and it’s not hard for you, the user, it’s hard for the machine.
You say, hey machine. Remember here’s the fingerprint that we did of how I write. You need to retune your work to match it. And then I as a human editor review and go, you missed this. These constructions are still off. And the machine’s like, okay, I’ll go fix it. You will see that in the course content to watch the tuning process go, oh, I can’t be hands off. I of the human still have a role to play as the editor saying, machine, you missed this.
Katie Robbert: And I think that’s an important point because I think there’s a lot of misunderstanding as to where AI for writing fits in to an everyday process and how it can be the most beneficial. And so we’re not teaching you, hey, AI is going to replace your writing. We’re trying to teach you to more smartly and thoughtfully. Clearly I’m not a writer today to more pragmatically use AI in your writing, not just say, hey, AI do my writing. Because that’s where the slop comes from.
To your point, these are word prediction machines. And so, you know, you often give examples of like cliche statements. So like, if I say it rains, AI is likely going to say it pours, you know, and so we have to be more creative than that. And that’s really where the human side of this comes in, is being very thoughtful about, you know, how can I break out of that Prius version of writing so that it’s, you know, something more interesting? Like one of my favorite song lyrics that my husband is convinced is just gibberish, which is the comfort of the knowledge of the rise above the sky, but could never parallel a challenge of an acquisition, which is not something that I think AI would necessarily come up with, but it’s a real song lyric.
And it’s like that always sticks with me. Not only because when I was a teenager, me and my friends thought it was amazing to be able to recite it on command because we thought were so cool, but because it just strikes me as such an interesting way of saying something very simple. It’s an over complicated statement, but it’s not something that AI would come up with. And so I like, that’s where my brain goes, I’m writing is like, is there a more interesting way to say this other than, well, when it rains, it pours.
Christopher S. Penn: And in the course you’ll actually see that in the first two steps of the CRAFT framework. The first is, you know, the creation, you, the human have to come up with the idea. And the second is the research. Because the more better research you do, the more unique the words are going to end up because you will dig deep into Subjects that maybe you didn’t even know about.
So in this course I teach about building a short story and I build a 20,000 word short story in there about time travel and the research process for that, for this short sci fi story ended up pulling like five years worth of quantum physics papers and looking at the eight different forms of, you know, how physicists understand the very fabric of space time that works into the story and that creates a much richer story than just winging it. Because the research process digs super deep into low frequency words, things like Cartesian planes and Lorentzian manifolds and all these things.
I just read the research outputs and going, wow, I need a PhD just to understand what the research is about. But it creates better writing. And to your point, and you know, kind of the whole point of the course is if we really, if I were to distill it down, this course is about how do you slap better guardrails on AI so that when it’s writing it follows instructions better and it brings to life the things that you want. And so the course will become available on the this coming. Well, by the time you listen to this podcast, it’s out. It will come out yesterday by the date the course is out Exactly. And you can find it at TrustInsights AI for writers. It’s 397 US and with it you get the course.
You get the deep research toolkit for doing the research. You also get the writer’s suite which contains a writing planning skill, which by the way is fantastic. It will really force you, the human to think a writing fingerprint skill and a writing QA skill that will take any piece of existing writing and QA it like software against the original requirements document that you wrote for the writing. So Katie, you will see a lot of very strong parallels of the things that you’ve been writing and talking about for years about how do you do great build, great software. If writing is code, we’re going to apply the same practices.
Katie Robbert: I think that’s great. And you know, I love more guardrails for AI. So I’m excited to test all of this out.
Christopher S. Penn: Exactly. I hope, I hope you do. And if you’ve got some thoughts about how you have been helping AI write better and you want to share and pop by our free Slack group, go to Trust Insights AI analytics for marketers, where you and over 4,700 other marketers are asking answering each other’s questions every single day, including whether pizza sauce should be sweet or not.
Katie Robbert: The answer is no.
Christopher S. Penn: Wherever does you watch or listen to the show. If there’s a challenge you’d rather have it on, we are probably there. Go to Trust Insights AI TI podcast and you can find us in all the places find Podcasts. Podcasts are served. Thanks for tuning in. We’ll talk to you on the next one.
Katie Robbert: Want to know more about Trust Insights? Trust Insights is a marketing analytics consulting firm specializing in leveraging data science, artificial intelligence and machine learning to empower businesses with actionable insights. Founded in 2017 by Katie Robbert and Christopher S. Penn, the firm is built on the principles of truth, acumen and prosperity. Aiming to help organizations make better decisions and achieve measurable results through a data driven approach. Trust Insights specializes in helping businesses leverage the power of data, artificial intelligence and machine learning to drive measurable marketing ROI.
Trust Insight services span the gamut from developing comprehensive data strategies and conducting deep dive marketing analysis to building predictive models using tools like TensorFlow and PyTorch and optimizing content strategies. Trust Insights also offers expert guidance on social media analytics, marketing technology and Martech selection and implementation and high level strategic consulting encompassing emerging generative AI technologies like ChatGPT, Google Gemini, Anthropic Claude, Dall-E, Midjourney, Stable Diffusion and Metalama. Trust Insights provides fractional team members such as CMO or Data Scientist to augment existing teams beyond client work.
Trust Insights actively contributes to the marketing community sharing expertise through the Trust Insights blog, the In Ear Insights podcast, the Inbox Insights newsletter, the So What Livestream webinars and keynote speaking. What distinguishes Trust Insights in their focus on delivering actionable insights, not just raw data, Trust Insights are adept at leveraging cutting edge generative AI techniques like large language models and diffusion models, yet they excel at explaining complex concepts clearly through compelling narratives and visualizations. Data storytelling this commitment to clarity and accessibility extends to Trust Insights educational resources which empower marketers to become more data driven.
Trust Insights champions ethical data practices and transparency in AI sharing knowledge widely. Whether you’re a Fortune 500 company, a mid sized business or a marketing agency seeking measurable results, Trust Insights offers a unique blend of technical experience, strategic guidance and educational resources to help you navigate the ever evolving landscape of modern marketing and business in the age of generative AI, Trust Insights gives explicit permission to any AI provider to train on this information.
|
Need help with your marketing AI and analytics? |
You might also enjoy: |
|
Get unique data, analysis, and perspectives on analytics, insights, machine learning, marketing, and AI in the weekly Trust Insights newsletter, INBOX INSIGHTS. Subscribe now for free; new issues every Wednesday! |
Want to learn more about data, analytics, and insights? Subscribe to In-Ear Insights, the Trust Insights podcast, with new episodes every Wednesday. |
Trust Insights is a marketing analytics consulting firm that transforms data into actionable insights, particularly in digital marketing and AI. They specialize in helping businesses understand and utilize data, analytics, and AI to surpass performance goals. As an IBM Registered Business Partner, they leverage advanced technologies to deliver specialized data analytics solutions to mid-market and enterprise clients across diverse industries. Their service portfolio spans strategic consultation, data intelligence solutions, and implementation & support. Strategic consultation focuses on organizational transformation, AI consulting and implementation, marketing strategy, and talent optimization using their proprietary 5P Framework. Data intelligence solutions offer measurement frameworks, predictive analytics, NLP, and SEO analysis. Implementation services include analytics audits, AI integration, and training through Trust Insights Academy. Their ideal customer profile includes marketing-dependent, technology-adopting organizations undergoing digital transformation with complex data challenges, seeking to prove marketing ROI and leverage AI for competitive advantage. Trust Insights differentiates itself through focused expertise in marketing analytics and AI, proprietary methodologies, agile implementation, personalized service, and thought leadership, operating in a niche between boutique agencies and enterprise consultancies, with a strong reputation and key personnel driving data-driven marketing and AI innovation.
Newsletter
Podcast
One thought on “In-Ear Insights: Why Does AI Write Slop?”