LLMs Got Everyone Sounding the Same Now

Hope y'all are having a good summer! Clearly, I've been spending much of mine continuing to reflect on the state of generative AI. Today, I want to talk about how the way we talk is starting to match the way LLMs "talk." I don't think that's good news for us long term. How the Internet Averaged Us One of the weirder consequences of the internet is that we have these giant shared spaces that connect us through viral content. We're all watching the same videos and reading the same posts because "the algorithm" has fed them to us. Sure, we have our pockets and echo chambers, but "the algorithm" is still feeding us more or less the same slop. Because of the shear volume of slop we consume daily, we're almost entirely desensitized to news. What's crazy today is forgotten about almost immediately. Like, I'm writing this between the time the Knicks won the NBA finals and their parade, and I've mostly moved on to thinking about this new Iran deal. Hell, SpaceX went public like five days ago. How can anyone keep up? What I find fascinating about this is that while we're being bombarded with story after story, we're also commentating on those stories. Everyone has an opinion, and everyone shares that opinion in, presumably, some unique way. You might post about the Knicks and then watch a few short-form commentary videos about the series. Over time, I feel like (as in I have no hard evidence) that this exchange has changed us linguistically. You have millennials like me who basically refuse to age because we're able to stay "hip," as my parents might say. I lean into words like "unc." Meanwhile, "bro," "bruh," and "brother" are basically my comfort words. Likewise, "mid," "glaze," "cooked," and "vibe" are everyday speech for me, and I even ironically use terms like "-pilled," "-maxxing," and "-rotting." Though, I still hold onto "haha" and "lmao" in text I suspect this is true for basically anyone with an online presence. You don't have to be terminally online for your language to change a bit. Like, it's even true for my dad. He started using popular phrases because he's watching the same videos as everyone else. In a way, I think everyone always being online has sort of averaged out our language. We're not really as unique and interesting as we used to be. Everyone is following trends, and fewer of us are are weird or different (unless that gets us views). We're All Just Talking Points What I find particularly problematic isn't really that everyone uses the same slang. That's mostly innocuous and kind of fun. What I find way more concerning is that everyone seems to share the same talking points. When I was ranting about the red/blue button debate, I kept running into this issue where people in real life were repeating the same arguments I'd seen online. Like, I repeatedly heard that phrase "I'm picking blue because I don't want to live in a world of people who picked red." Like, it's pithy and clever, but I know you didn't magically surface that idea on your own. It's literally the same argument I've seen espoused like dozens of times in comment sections and videos. Side Note: since I've been gaming a lot lately, I've been thinking about how games that are built around metas become really annoying when the metas become stale. For example, you might play a game like World of Warcraft where certain builds are better than others. After each patch, I have to imagine the game is more fun when people can't immediately min/max their build through Icy Veins. You can't get flamed from a teammate for not running the best build when no one knows what it is. That's kind of how I feel about these online debates. When everyone already has their talking points preloaded, there isn't a lot of interesting discussion to be had. It's not that I'm expecting people to have novel ideas. It's just that I can tell that no one has really thought about the question themselves. It wasn't like the button problem was presented to them in a vacuum. They simply saw a video or post from a creator that they like, and they repeated their talking points verbatim. I feel like this happens with everything now. You sit down with your conservative parent to discuss politics, and they're incapable of engaging with the ideas. They're simply parroting whatever they've heard from whatever they consume. Like I literally ran through this recently with one of my loved ones where they kept repeating silly phrases like "ICE is only going after criminals" or "Islam is a dangerous religion." We're all just talking points. Case Study: Social Media I bring this all up because as much as I try to stay offline, I find myself crawling back to social media every once in a while. You might know that I left Facebook ages ago and more recently I left Twitter. In the interim, I was using Reddit anonymously to get my fix, but that site is so hostile to browser users, especially on mobile, that I've been weaned off naturally. So, you might be wondering where I could possibly be now? Is it Bluesky? Maybe Mastadon? How about LinkedIn? Nope! None of the above. I actually had the genius idea to get my news from Substack. What a horrible choice that was. Substack is insufferable in a sort of unique way. While Facebook is full of racist boomers and Twitter is full of Nazis, Substack is full of smarmy writers and academics. It's the kind of place you might go to if you want every point you make interpreted in bad faith, kinda of like how Twitter was before Elon. Except, there's less dunking and more walls of text. Of course, Substack is relevant to this piece because it's a really interesting case study on what happens when a social media platform for writers embraces AI. What you end up with is a natural schism between writers who hate AI and "writers" who don't. And, incidentally, all the the "writers" who embrace AI all sound exactly the same. Let me give you an example. I was recently following the AI discourse on Substack when I stumbled upon an article titled, "The many fallacies of 'AI won't take your job, but someone using AI will'." It's an "old" piece by internet standards, published in April 2025, but I thought the title was provocative. So, I dug a bit deeper. Right away, I was really weirded out by the comments. After all, the comment sections under most posts on Substack are fairly hostile. I've seen so many "debates" around arguments that authors never made, and this article had none of that. It was almost bot-like in the way that every comment was like "wow! So insightful! I have preordered your book." Some of the comments even read like AI. Like, do real people talk like this? The Maginot Line metaphor is overused, yes—but here it works because you don't just gesture at it, you commit to it. And in doing so, you illustrate exactly how these aphorisms function: not as clarifying insights, but as consensus sedatives. There's just no way these are real people, right? Absolutely loved this. As a filmmaker working with AI, I see how the disruption isn’t just about tasks being automated — it’s about the deeper restructuring of the creative process, authorship, and the systems we’ve long worked within. Just the production phases we’ve always known are broken down completely. We’re not just adapting to new tools; we’re navigating a shift in process, output, power, narrative, and value. I'm not even bothered by the em-dash here. I love those guys. It's the parallel construction that is repeated over and over again: "it's not X, it's Y." It drives me absolutely insane, and I refuse to believe that people are consciously writing like this repeatedly. Like after the second time you do it, you have to imagine it starts to feel repetitive. Naturally, given how many people sound the same, I started to wonder if the author was also a bot. So, I took a peek at their article, and it sounds just like the comments. AI doesn’t simply replace tasks. It reshapes the architecture of the system in which those tasks used to make sense. The arrival of the intermodal shipping container did not simply automate loading and unloading of ships. It restructured the economic logic of trade, making some ports like Liverpool irrelevant and changing the fortunes of others like Singapore. Not because they are done poorly, but because they no longer create any advantage. In any system, value accrues not to the most productive participant, but to the one controlling the scarcest complementary asset While the article is littered with this same writing structure, it also contains far too many niche examples. Like, it starts with the Maginot Line. Then, it talks about dock workers, fashion, basketball, typists, longbowman, cricket, excel, and music streaming. I get it's a long article, over 4,000 words, but there is an absurdity to the number of examples which are covered with almost no depth. Bro said "give me a list post that deconstructs the mantra, 'AI won't take your job, but someone using AI will,' and include gripping examples for each topic." Given all the telltale signs of slop, I was inclined to question whether or not they had written it. Which is a bit troubling given that their bio reads: "Author of multiple best-selling books." As a result, I decided to check out their books to see what people were saying about them. First up, "Reshuffle," which sits at a 4.6 on Amazon. The top reviews kind of all read like the Substack comments, but they are nonetheless "verified purchasers." Like, be so for real: This book feels less like theory and more like a roadmap — one that left me reflecting on my own work and excited about the opportunities ahead. Highly recommend, especially if you’re looking for fresh ways to think about leadership in the age of AI. Reshuffle is not another “how to use AI” guide. It’s a powerful, big-picture look at how AI is reshaping the very foundations of the knowledge economy. Sangeet doesn’t just explore tools—he reveals the tectonic shifts in how knowledge is created, distributed, and valued. Most people use AI to improve old systems; this book shows why the winners will be those who understand and adapt to entirely new ones. Using powerful examples from history, like the bar code, container boxes and the Maginot Line, Sangeet creates powerful frames for new ways of thinking. Insightful, clear, and compelling, Reshuffle is essential reading for anyone who wants to lead in the age of AI This kind of stuff makes it look like there's not a real person in the bunch, just a bunch of paid actors. What I found much funnier were the critical reviews, like this two-star review: I really wanted to like this book. It talks about how AI will help coordinate data and revolutionize workflows. Great - i want that too! But it only keeps repeating that "theory" and never gives any practical advice or how to info. In fact, I suspect this was actually written 90% with AI. I have to wonder if there are people out their building out their entire personas using bot farms. Like, I had never heard of this person, but they have a Wikipedia page and multiple best-sellers. Yet, I can't find an ounce of criticism of them online. Like, surely people have to hate them, right? Meanwhile, basically any interview he's in on YouTube is lost to the algorithm; no comments or views. It's giving grifter. Accusing People of Using AI Is Not the Point I'm not necessarily accusing any of these folks of using AI, even if the article contains a "comic strip" with the patented piss filter or even if the "author" somehow managed to write the wrong publishing date of their book. After all, I don't love the idea of hurling accusations of plagiarism without evidence. I think I ranted about this elsewhere on the site, but I think we could learn a lot about the damage that these kinds of accusations can cause from the chess community. After all, we'd still have Naroditsky with us otherwise. For the record (and apologies for the massive tangent), I'm really torn on this issue. On one hand, I really don’t care for witch hunts or purity tests. It’s how you ruin the lives of innocent people. On the other hand, I think everyone knows that I believe generative AI is a unique evil (not in the becoming sentient sense, of course; don't buy into the AGI hype). I don't want my tattoo artist to use AI, even to prototype my ideas. I don't want my authors to brainstorm, draft, or edit their books with AI. I don't even want AI assets in video games or AI voice lines.I'm not being a selfish consumer when I say these things. I'm not claiming ownership of the works of others. I'm simply stating that I personally don't want to consume media that was created with AI, and I feel like the people that use AI want to trick me into consuming it. "See, you couldn't tell. It must be good art, right?" It's weird. Why are we excited about being tricked by machines? And because of this deception, it's only natural that we might be hesitant to enjoy media. How could we if we're constantly at the risk of falling for the bait? We're simply exercising caution, which might look like purity testing from the outside. If enough signals point toward AI—the piss filter, the extra fingers, the em-dashes, the lack of continuity, the "not x, but y" pattern, the hallucinated sources, and even the creator's positive disposition toward AI—we're going to call it out. I think that's the only fair way to react when we're always at risk of being deceived. Is there a correct way to go about calling someone out for using AI? I don't know. Again, I don't think it's worth destroying someone's reputation without undeniable proof, but I also don't think people using AI should consider themselves artists, musicians, writers, or even developers. Yes, there are ways to use the tools more sparingly, but I'm a purist on this issue. Throw that shit in the garbage. But in keeping with the theme of this article, I find it troubling that I can't distinguish LLM output from human writing, and I don't think that's because LLMs have suddenly mastered language. Rather, I think that because people are talking to chat bots, using chat bot output to compose emails, scripts, and articles, and consuming media created by chat bots, people are starting to sound like chat bots. I'm sure there's already been a ton of linguistic work on this, like people are unironically using the word "delve" now. That seems bad to me long term. The internet already has 60-year-olds sounding like zoomers. Now, ChatGPT has us all sounding like bots. I don't really know what to leave you off with at this point, but I'll just say that it might be a good idea to start touching grass. That's perhaps more advice to myself than you because I really need to stay off these slop sites. They make me feel insane. If you've not quite been driven insane yet, maybe you'd like to read a bit more of what I have to say: The Future of AI Is Ubiquitous Surveillance What Happens When I’m Forced to Teach AI? Missing the Forest for the Trees: Why You Struggle to Solve Problems Hell, if you really enjoyed this, consider helping me grow the site. This site is entirely self-funded, and I won't be shilling to private equity any time soon. If neither of that is for you, no worries! I hope to see you again. I wanted to jump back in here before this thing publishes to share a video I saw recently that aligns with a lot of what I said above. It comes from @etymology_nerd on YouTube, where they talk about how AI is changing the way we speak. It's really short and briefly cites some of the current research on the topic (though, the one source is a preprint on arXiv). In short, it basically reinforces a lot of the arguments I make here, which I feel like are obvious. But, I'm happy to have some empirical evidence.