Everyone Sounds Like ChatGPT Now (Including You)

AI slop is literally everywhere right now. It's, it's in all the news feeds, social media, YouTube videos. It's on TV scripts that are being written for people for their speeches, even like best man speeches. It's like AI is being used for everything. So it has us all questioning. It's like, what is real and what's not real? The lines are being blurred and our thoughts are kind of not becoming ours anymore. You know, when we're also reading from ChatGPT and Claude and Gemini and all these other things all the time through the work that we're doing, we're just going to just start picking up some of these Claude-isms and ChatGPT-isms, and then they're just naturally going to kind of leak out into what we're saying and writing ourselves without us knowing. And that's kind of trippy if you think about it, because our minds are kind of at risk of being hijacked if we're not aware and checking ourselves. There's a time that, which is right now, I think that we could do something about this and You know, we could prevent our, our thoughts from kind of slipping away from us before it's too late. And there's actually a guy that I brought onto this video that knows exactly how this all works, and he's doing something right now about that.

It's getting harder and harder to realize what is actually coming from my own brain and what was coming from, you know, a discussion that I had with an LLM recently. So much so, that I've literally stopped using AI to like brainstorm ideas. I have my old-fashioned handwritten notebook that I literally go outside—

Yeah.

With my coffee every single morning and take notes while sitting there with a, with a coffee to just like unleash my thoughts into a notebook. Some areas where I will get help with videos is I will have LLMs help me outline a video, right? I will turn on a microphone and say, hey, I wanna make a video about What should be automated with AI versus what shouldn't be automated with AI. And here's my thoughts. I'll brain dump for 20, 25 minutes into one of these things and say, this is the content of the video. I wanna make this video. Don't change my thoughts. Don't add any extra information. And now I have an outline, but it's not writing a script for me. It's not the AI deciding what I'm gonna talk about. It's not the AI really doing anything. Other than taking my raw input and saying, put it into this order.

So that's Matt. He's awesome. Known him forever. And he's deeper down the AI rabbit hole than anyone else I know and anyone else I follow online. I mean, I've known the guy for over 20 years and the guy lives and breathes this stuff. What's interesting is I didn't realize, but in this video I had AI help Outline this video for me. Like, the show flow is not AI's words at all. I brain dumped everything into one of my recorders. I use Granola. And then I put my thoughts, my raw thoughts into Claude to just make an outline for me. So it's just interesting that Matt literally is talking about doing the same thing with AI, use it as a tool to shape things around, but not actually change the actual words. that you roll out. AI is so integrated, I didn't even think about it. It's just like what I do for my work and workflow. And I think that's kind of the power. And also, you know, like the scary part of AI is like a lot of this stuff is just becoming so second nature for, for us that it's gonna start slipping into all these other things that we do as well without us knowing.

I remember there was a, a, a tweet that I was replying to and somebody had this like big breakdown of a tweet and I started to reply to the tweet. And my reply started with, what he just said was directionally correct, but not completely correct. And I went, when have I ever said directionally correct before? And then I realized LLMs say that you're directionally correct all the time. I had never said that in the past since using ChatGPT. And it was like, as I was writing this tweet, I sort of started to write, you're directionally correct. And then I like read my tweet back and went, Anybody who reads this is going to think an LLM wrote this, even though I literally did not use an LLM to write this right now. Wow.

Directionally correct. Yeah, I know. I mean, like, there's certain, definitely like phrases or delve, obviously the whole like em dash thing. I feel like people have like figured out how to cut that out of their AI responses as a whole.

Yeah, I would say the biggest pattern, like if you are looking for patterns in this kind of stuff, is the, it's not this, it's that kind of pattern, right?

Mm-hmm.

Like, this isn't only a great idea, it's, uh, foundational to what we need to learn right now. You know, like it, they, they have this pattern of, it's not just this, it's this. And whenever you see that, it's almost always AI written. And I'm not saying like there aren't people that spoke like that before AI, but now AI writes like that so frequently that That to me is the biggest tell. I always immediately assume it's an LLM.

Why do you feel like this is happening where it's just kind of coming out like that wasn't the AI speaking and, and, you know, writing that, that tweet response, that was you?

I just think that people are building too much of a relationship with AI. I think people are having conversations with AI, talking to 'em like it's their buddy, like it's their, uh, you know, their personal assistant, their, uh, in some cases their, uh, significant other, right? We've all heard stories, right? And I think, I think people are having these like, conversations where they're going deep and they're having really deep conversations. You and I talk all the time, right? You and I were business partners. We had a podcast for years and years and years together. Uh, we're starting to do live streams together, all of this kind of stuff. You and I talk frequently. What happens when you and I talk frequently? You start to adopt some of the things, the ways I speak.

Sure.

I start to adopt some of the ways you speak. We adopt mannerisms. Like, stuff sort of blurs together when you have conversations with people often. I think that same thing is happening with AI. I think people are just having these deep conversations with ChatGPT and the next thing they know, like the way ChatGPT speaks is just embedded in their brains and that's how they start to have conversations with other people. And I don't think that's necessarily a good thing, right? I think as I learn more about the fact that, you know, back when we were using it 6 years ago, we didn't know how that data was trained. We didn't— Not at all. Understand they scraped the internet and basically said, this is all ours now. We're, and, and now we're gonna sell it back to you, right? We didn't know that, uh, they were going to need to build these mega data centers that were gonna have to be built close to communities and create extra noise pollution and, you know, siphon off water from communities and use extra electricity, which could cause electricity costs to rise in these communities. We didn't foresee the rise of AI slop and everybody just generating AI videos and making advertising online of stuff that doesn't actually exist. And scamming people. Like, none of that was on our radar when we were playing with this in the early days. And as we've learned more about that kind of stuff, I've course corrected around that too. Like, I'm much less of a fan of AI video generators than I was early on.

So Matt is no AI skeptic. He's been reporting on the AI news and going deep down that rabbit hole for something like 5 years. Before that, him and I were using AI in a business that we were running together. So it's, it's interesting that he goes to pen and paper and gets away from tech to capture all of his thoughts.

Typically I, I make a cup of coffee, I, uh, grab my notebook, my pen, and I go and sit outside with a pen. And I don't know if you could see this in the background, but I have like all these different highlighters here.

Yeah.

And like, I actually have a system. I just go and like sort of brain dump whatever's on my head. But like different highlighter colors mean different things in my notebook. Like if I highlight it in pink, It's a video idea. If I highlight it in green, it's something that I want to keep on coming back to and I review all the time. If I highlight it in orange, it means I don't need to review it again. It's like maybe that was a to-do task or something that was on my mind that, you know, I, it's no longer on my mind anymore. So I highlight it in orange and my brain knows to skip over anything that's highlighted in orange. I highlight it in blue if it's an app idea and I highlight it in yellow if it's notes from like a keynote session. so that I can quickly flip through my notebook and because of the color coding, find exactly what I'm looking for. I do have a thing I do every day that I, I, that are like the 3 most impactful things I can do today.

Hmm.

And I always write a list of like at least 3 things that I wanna do today that will have the move the needle the most on either my content or my business or, you know, something like that. I'm not getting those 3 impactful things from an LLM. I'm getting 'em from my brain. I'm getting 'em from whatever was on my mind As I went to bed last night or whatever.

Well, yeah, they're, they're in your mind, subconscious or wherever. And then you're, you're removing yourself from all these other inputs that not only LLMs, but obviously your phone has all these reminders or other people on Slack saying, hey, we gotta get, like, this is my priority now, it's yours kind of things. And I gotta give you credit, like, you've been good at that for a long time.

I just love systems. Like, if I can build a system around something, I will do it and I will stick with it for a while. And then my systems always shift as new technology or new ideas come along, but like, I always have systems and I'm always trying to build systems around everything I can think of.

To that point, what you just said is like technology shifts. So that means systems either get upgraded or they tweak a little bit, but it seems like the paper thing, like what you just said, is a personal system that is kind of more foundational than anything, right? Like, because that can then extend into whatever technology's doing at its time.

To me, there's just something about pen and paper that is still very anti-technology and still very sort of like raw. And I'm able to unload my thoughts there better than anywhere else. I also find pen and paper is the easiest possible way to review notes, right? Like if you're using something like a Notion, I find it much more difficult to go back through your archives and review old notes where when I have a notepad, like I go through and highlight stuff in green of like, you need to review this constantly. And I go and sort of review my notes almost every single morning. So it's like, for me, it's really, really easy to go back through the archives and constantly get these reminders of things that I was thinking about 6 months ago that I still wanna be on my mind today.

So what Matt is doing is he's taking his thoughts, he's capturing it on pen and paper and also in audio form or even typing form, and then taking what he has captured himself, his own thoughts, and then he puts it into AI. And then has that structure his thoughts around into maybe a format he's looking to do, but it's never diluting what he says. His words are the inputs and it's always the first thing that goes into AI and then he's shaping it around, but it's not changing what Matt has said. That's the context. That's like the foundation of everything. Never the other way around. It's always inputting with his own words and you can do the same.

One actually good thing to do is like when you do that sort of journaling process, Take that journal entry that you just wrote, give it to ChatGPT, and tell it, find the gaps in my logic. Tell me the areas where I'm thinking about this wrong. Like, don't ask it to validate your journal entry. Ask it to tell you the holes. Ask it to tell you like where you're missing things. Ask it to tell you like the, the flaws in the logic of what you just wrote up. That to me is way more helpful than just like, Here's a journal entry. What do you think?

Is this where you would have AI kind of structure things like, let's say for one of your videos or something? Yeah.

And everything we've talked about so far is not necessarily for the purpose of making videos. Right. You know, maybe, maybe getting both sides of the perspective on a piece of news so that I can talk about it intelligently. That's more sort of like me just sort of reading up on it. That's like part of the research process. That's just like, here's the news. What, what are all the considerations I need to think about when I start talking about this news? But it's not necessarily gonna turn into a script. It's just kind of like implanting both sides of the token into my brain. Mm-hmm. So that when I do talk about it off the cuff, both sides are up there, right? Yeah. Both sides live up, up in my head. Some areas where I will get help with videos is I will have LLMs help me outline a video, right? Never the script, but sometimes an outline. And the way I typically do that is I will, do the sort of unloading process, right? I will, Turn on a microphone and say, hey, I wanna make a video about what should be automated with AI versus what shouldn't be automated with AI. And here's my thoughts. I'll brain dump for 20, 25 minutes into one of these things and say, this is the content of the video. I wanna make this video. Don't change my thoughts. Don't add any extra information. Just outline this so it's in more of a logical order. If I just brain dump that into a video, it probably wouldn't be a logical flow. Help me rearrange these talking points into a flow that feels more logical for a video. And now I have an outline that I'll keep to the side of my screen. And as I'm recording my video, I'll be like, okay, I mentioned that. What else did I want to mention? Okay. And then I'll go back to my recording and I'll start just riffing and ranting on the next point. And I'll be like, okay, I made that point. What's the next point I'm supposed to make? Okay, that's the next point. And I'll start riffing and ranting on the next point, right? So like I'll have an outline of points that I wanna riff on, but it's not writing a script for me. It's not the AI deciding what I'm gonna talk about. It's not the AI really doing anything other than taking my raw input and saying, put it into this order.

Is there like a hard refusal of things like just quickly you can like riff out that you would never have it do?

Scripting. I'll never let it write my scripts. I, I don't know if I wanna say never for thumbnails, but it still feels so far off at generating good thumbnails.

Mm-hmm.

It's decent at making like elements of a thumbnail, right? Like let's say I want a monkey on roller skates in my thumbnail. Like that's some, an example I used to use in videos all the time.

Right.

I can have it generate an image of a monkey on roller skates, but I'll still work with my thumbnail designers to, you know, in Photoshop make a really good composition and maybe one of the images in the thumbnail is like that monkey on roller skates, right? So we'll use AI to make like elements of the thumbnail, but trying to get AI to generate thumbnails, it's still really, really bad at that. So I don't think I would do that. I will never, ever, mark my word, never, ever have it write social media posts for me. That is the cringiest shit I've ever seen, is when I get on Twitter or LinkedIn and you read an article or a post and you're just like, this is something ChatGPT wrote, and you're trying to pass it off as you being smart. But anybody who knows what LLMs write like, know that this is an LLM. Social media is meant to be social. It's meant to be your thoughts from your brain. I will not let AI write social media posts for me. It's hard to say never on like anything, but I know I don't want to use it for scripts. I know I don't want to use it to generate full thumbnails. I know I don't want to use it for any social media posts. I want people to know that if they're seeing something on X from me or LinkedIn from me or one of these platforms, it was actually written by me.

It's basically anything where your voice is being published outward, right? To other people.

Yeah.

And, and obviously thumbnails too, which is like, you know, that's more of like a quality thing, it sounds like.

I just think fully YouTube-generated thumbnails are really cringy, and I used to make fully YouTube-generated thumbnails in the early days, but back when I did it, I was like the only person doing it, so it was a different era. Now I see it and I'm just like, ugh, why are people doing that?

And right now more than ever, We are the creators. Everybody's a creator. You're a creator. And what we publish is what the world sees as us. That's what's real. So it's up to us to publish what is most real in our thoughts and thinking, what discernments we have about what's being discussed instead of the average, and then making everything we put out AI slop without our discretion and our actual thoughts behind it. So that's the big thing here, is knowing that you are a creator and we have the access to amazing tools that have never existed before, and literally they're free. So we have the ability to publish some amazing things that we never did before, but it's up to us to capture our own way of thinking first and then use the tools in a way that we can actually go out there and be the powerful creators that we all are.

When it comes to being a content creator, there should be more reason to have the video out there other than to just have the video exist, right? Like, anybody can go and pull up information about a topic, throw it into ChatGPT and say, write a script for me. Take that script, throw it into ElevenLabs, have ElevenLabs speak out the entire script, And then go and take that video and have, you know, uh, Claude, uh, create B-roll over that entire video, right? You can have entirely AI-generated videos where it's informational and it gets some sort of point across and pretty much no human touch was involved other than maybe typing a few prompts. What is the reason that video exists? Because if that information is freely available, like, 100 people can make that exact same identical video and, uh, and send that same information out there, right? Like, I think what makes videos like mine different is that you are getting a human voice. You're getting somebody with nuance, right? Like, I'm somebody who's been focused on AI for, you know, over 5 years now. And I'm also out there going, there's a lot of shit about AI that sucks and we shouldn't be doing it. Most AI channels aren't going to have that nuance.

Yeah.

And I think that, that, that commentary, that nuance, that human voice of like, look, there's a lot of cool things about AI, but also a lot of really bad things about AI and the stuff that most people hate about AI. I also hate most of those things about AI. And like, I'm gonna keep on telling you what's going on in the AI world, but I'm also gonna share my thoughts on what I think is good about it and what's bad about it. And, um, you know, you take that humanity out of some of the videos, you take that nuance outta the videos. If you're just making something that's like informational with B-roll over it, you're literally making something that anybody on the planet can make with some prompts. Now there's nothing that makes it unique, nothing that makes it interesting to anybody else, and the internet is gonna get flooded with that kind of stuff. So it's gonna be commodity content and it's just like, I don't get the point.

All right, so this week, think of something that you wanna publish. So if it's working through a bunch of emails, maybe writing something, a document, presentation, speech, you know, a video like this, whatever it might be, capture your thoughts first. Like just capture all your raw thoughts in written form or maybe in audio form. And then you take those raw thoughts of yours into the LLMs. So go take it to ChatGPT or Claude or Gemini and then prompt it to get some structure to it. Don't actually have it do the scripting, but flow it and arrange it in a way that makes sense for what you're trying to to deliver. Prompt the AI to spot any gaps or any counterarguments or anything that you might be missing. If it does expose some gaps or some things that you might have missed, that's when you can layer in more of your thoughts right to the LLM. Just type it or chat right to it, and then eventually, you know, get to a point where it's ready to, you know, it's usable. Now you can go draft that email. And I would not suggest just have AI start to write all your emails or do your or scripting like we talked about, but now you have a flow and you can work with it and make it your own. You know, there's still this human element that we gotta keep alive and have our thoughts actually become things in the real world and not just defer all this stuff to AI. You know, with the world speeding up faster than any of us can keep up with, we need to keep ourselves in check and, and make sure we're capturing our thoughts and doing something with it. If you like everything that we talked about and if you want more of this, please like, subscribe, leave a comment on your thoughts. Maybe there's another topic that you want me to break down. I'm happy to do so. Thanks for watching. I appreciate you so much and I'll see you later.