AI Is Making Your Thinking Worse (And How to Fix It)

I really did not want to make this video, but my team was not using AI correctly to the point where like I had to write an entire memo about it. This is the memo that I wrote. So many of you guys are using AI in ways that make you worse and you do not even realize it, and the thing is is that people can tell when you're faking it and when it's f***ing AI. It is not a good look. So if you do not want to get embarrassed by AI, there are three things you should stop doing, and there's one thing I will beg you to start doing. Number one mistake is that you are letting AI think for you. People think they're using AI to save time, but what they're actually doing is skipping the most valuable part of their work. Say you put this in chat, you say, write an email to my boss about my team not working together. Your boss gets the email, then asks you questions about it. There's a lot of questions that they ask you that you realize you don't have the answer to. Why is that? Because you haven't actually thought through the problem clearly enough so that somebody else can actually help you solve it. What happened? What have you tried? Have you thought about other ways to solve this? What do you need from the other person? The process of writing something and thinking through it to write it clearly is what is so helpful about writing in general. There's such a low barrier to entry because you can just ask ChatGP to write something for you that's generic. What it means is that it is stolen from you the biggest piece of value that you get from writing, which is that you no longer think. And so you actually don't understand the problem well enough to stop creating it or to even solve it. Recently, it was probably about four weeks ago, somebody from one of my departments wrote me a memo about a problem that they had, and it was clearly all written by AI. And it was essentially a proposition for a new philosophy that we would have in the department. Two things that were wrong with it is one, this memo was like eight pages long. And I want to say that it probably could have been a half a page because it was just full of fluff and bull. The second thing that was wrong with it is that it didn't name a specific solution for our company. It named a very generic solution that still needed to have a solution within it. It's like, we need to come up with a philosophy for this thing. Okay, sure, we need a philosophy, but we also need to solve the problem. Now, why is that? Because AI, when you don't give enough context, which most people are pretty lazy at prompting it, will just give you generic answers. It has to level up and chunk up on the problem because it doesn't have enough context for you to chunk down. People are so lazy using it, they don't even prompt it correctly. They don't take the time to write four paragraphs of context to even get a somewhat useful answer out of it. And so what I've realized recently is like, I would rather read a rough draft or handwritten from a pigeon with real thinking behind it than some polished, very long, eight page memo that says literally nothing because this is not helpful. It doesn't help us actually solve the problem. It just looks like work. And this is what people miss. The value of writing was never in the writing itself. It was in the process required to write something. I've learned this in the fact that like, I just got done with my manuscript of my first book and the process of writing, it clarifies my thinking. I then understand the problems and solutions to a level of depth I could not explain to you. And if I had used AI to write my book, I would not have gained that. And therefore I could go on all these interviews and podcasts. I could do a book tour. I wouldn't know what to say because I wouldn't understand the problem deeply enough. So if you outsource the thinking, you're not actually getting leverage. You're giving away the exact part that makes you valuable. So before you use AI, ask yourself, have I actually thought through this myself? Am I using AI to organize my thinking or to replace my thinking? And I would argue that you want to start with organizing your thinking, not replacing your thinking. There's an easy way to tell when someone hasn't done the thinking yet. The writing has like a very tangible pattern to it. This brings me to number two, bad AI is not subtle. Okay. Most AI writing, especially if you're using Chachi Petit, which please for the love of God, use Claude, not Chachi Petit if you're going to write. Most AI writing sounds generic. Like somebody is saying all these fluffy filler words instead of actually being smart. So I can instantly tell that AI wrote your email or your memo when you use these phases. It's like the second I see something that's like in today's rapidly changing landscape, or as we navigate the complexities of, or, you know, the synergies of the, what's another one? Signal. Like, oh my God, signal? Who the says signal? Nobody says these words. I don't know who, by the way, programmed it to write like this, but it's awful. And you should change it. The other day I wrote about a one page memo and then I put it into AI just to organize it. You know, I just want to organize it into sections. So it's a little bit easier for people to read. And I remember I put it in to AI and I'm not even joking. Every single paragraph, I told it not to edit it. And it still did. It came out with like in the rapidly changing moreover, furthermore importantly. And I was like, why would it do this? And then the funny thing is it had this little like ending where it was like, at the end of the day, it doesn't come down to this or that. It comes down to, and I was like, oh my God, it is just so many wasted words and such a waste of people's time to read. What it turns into is like, that is AI slop. Literally two days ago, I had a leader send me, they're like, Hey, I was going to send this to the full team. And I was like, this is written by AI. And they were like, what? I was like, dude, you don't talk like that. The reason this stuff is so frustrating is not just that it is stylistically bad. It's because it reveals something deeper about people, which is that nobody made a clear decision about what they were trying to say. And nobody thought to themselves, I value the reader. So I should make this clear and concise. And when you think about good communication, good communication is being clear. Most of what people write with AI is very, very unclear. And so instead of what I would say is like, you have to decide what is the point of what you are writing? What do I want to have happen from this person reading this memo, message, letter, substack that I just wrote? How do you want them to feel? What do you want them to do? Do you want them to take action? If you do use AI to write, which I would say, I would argue for you not to do. But if you do, read your drafts out loud. If you wouldn't say it, don't write it. This is actually one of the best things I've learned through writing my book. One of my forms of editing has been that I read my book out loud. So twice now I've read my entire book out loud and asked myself, I'm like, oh, I wouldn't say it like this. And I didn't use AI to write. I f***ing wrote it myself. But sometimes you write differently than you speak. So imagine how different it is when you're actually using AI. Bad phrasing is just a symptom. The real problem is what is underneath it. Number three is this pattern matching. AI makes the writing better because it makes it longer or it sounds nicer or it sounds like pretty. But I don't think that things sounding pretty is the same as things being clear. Clear writing comes from clear thinking and clear thinking comes from clear writing. It's like a chicken or egg problem. These all come from having good judgment. What that means is that you have to use your brain and think. You have to decide what happened, what matters next, what changed, what am I going to do about it? What does it mean for my team? Now, AI can mimic language and patterns and such and maybe even decisions for you. But that is very dangerous because at the end of the day is you start realizing that you have used AI to solve problems. That somebody else brought the problem to you with AI and now you're going to send somebody your AI solution and they're going to use their AI to review it. And what does that mean? Do you think anybody is actually solving the problem? Probably not. So here's the rule. If you've written something like you can't argue for why that paragraph matters, f***ing delete it. One of the biggest things I've learned to write in my book is like the power of deletion. Each word on a page needs to like fight for its life. Let alone a f***ing sentence or paragraph. So if it's so unclear that even you cannot explain it, you are not ready to send that, to publish it, to put it on f***ing Twitter. Dear God, don't put it on Twitter. Don't put it on LinkedIn. Just keep it to yourself. If the answer isn't getting AI to just like generate all this s*** for you, what are you actually doing instead? Do not be afraid to get messy. If you want your writing to sound human, the material has to come from human. For example, guess how I wrote the notes for this video? I didn't type it and I didn't use f***ing AI. I wrote it with my voice using WhisperFlow, which I f***ing love that app. And then afterwards we go and edit it and format it. And so essentially what it is, it's the tool is AI voice dictation. So it's like very good in the sense that it can even hear the nuance to your voice. So it understands if you're asking a question, you need an explanation mark, you need a period here. And so you just say it into the app and then it can write it for you. And then from there, it also has punctuation and grammar, all those things that typically we don't want to do when we're typing. And then the funny thing is it sounds like a human after you use it because it is a human. You have done it. You've used your voice. It is much faster than typing it. It's not cheapening your writing the way that AI does. So for example, for me, I spoke into WhisperFlow to write my book. Why was that so helpful? Because one, I don't think using AI to write a book, especially one that's supposed to be meaningful and represent my brand would have been a good idea. But two, it forced me to think through things. And I think in twofold, which is like, one, if it doesn't feel good coming out of your mouth, I don't want it in writing. So it's like you immediately kind of gauge yourself when you're speaking. I think the second thing is that what I've realized is I can gauge when I'm overexplaining or I'm going too long. And so it's actually been a very helpful tool to use. If you bypass that process, you're bypassing your own thinking, which is why I have you here. I would have agents and robots if I didn't want your brain. We could just replace ourselves with them if we don't want to use our brains. And so I think the idea of giving ourselves the space to think through a problem enough, those are all things that we don't do if we're just asking AI to write something for us. And that process can only come through you doing it yourself, trying, failing, seeing what works, and then, yeah, maybe feeling self-conscious, like, I don't know if this is written well. Who gives a f**k? At least it's written clearly or it's written in a way that exemplifies what you actually think. I really believe that AI is at its best when it has real substantial content to work with, having your opinions, your logic, your phrasing, your arguments, your thinking, and then you can put it in AI to maybe ask it some questions and prompts about that. But if you start with nothing and you ask it to generate the whole thing, you do not get the value of solving the problem. You just get AI slump. Do not start a prompt in AI with, write me an email about this. Start with your own first pass first and use it to help you iterate and organize. That is the right use of AI, especially right now. Because once you've done the thinking for yourself, then AI actually has something useful to work with. So what does that mean? What do you actually do? Okay, once you have a draft, that is when AI can be actually really useful. What I love to do is I write something and then I will put it in AI that has context on my life and my business, etc. And I will ask it things like, hey, if I were an employee reading this in my company based on everything you know about my company, what questions would this reader still have? Where is my language unclear? What fears would this provoke in people if they were reading it? Does it seem like I have a logical argument? Do I say anything in here that would offend anybody? And then sometimes I'll tell it what I would like is for them to walk away and feel that what I've decided is fair. Do you think that I present my argument in a way that would portray that to the reader? So I think of it more like an editor who has to read everything but doesn't have the context I have. So it can spot holes in the logic, but it can't actually give me all the answers of how to fill the holes. Does that make sense? So for example, when I have been writing my book after I finish a chapter, I have a whole story in there and it leads into a point. And then one of the most helpful things I've asked of AI is I'll say, okay, hey, here's my chapter and it has context on my whole book, everything like that I've already loaded in the context window. You know, one of the first questions I'll ask it is, where is this repetitive? Where am I repeating points in this chapter that I've already made to the reader that are just probably losing them? And it will highlight, here's the points where I think you might be losing the reader. And I'll say, okay, great. The goal of this chapter is X, Y, and Z. Is there any part of this chapter that is going against my goal? Are there any side quests? Am I detracting or distracting my reader in any way by talking about these other things? Really helpful questions to ask. The cool thing is that AI will give me suggestions. I get to decide. And there's many times where it says, like, I think you should cut this out. I'm like, I don't think so. Like, I'm not taking it as the word of God, taking it like an opinion, like somebody else, if I were talking to somebody, ask them for an opinion, I would say, great, I'm gonna take that with a grain of salt, which is how you should all think about AI. You should not think about it like it is final and fact. You should think about it like it is a opinion to be considered. AI should be your editor. I would say possibly red chair in terms of like when you make decisions, which means like challenging your decision. It should be an organizer. It should not be your ghostwriter or your decision maker for your company or for any problems in your life or responses that you need to give people like the people who get the most out of AI are not the ones who use it to start things from scratch. They protect the parts that they know they should be doing, and they use it to augment the rest and get leverage on all the other work that has to get done around that. Here are the steps I would suggest. One, you should think through the problems first. Do not outsource them to AI. Two, use whisper flow or some sort of voice talk to text to draft your thoughts. Third, then use AI as an editor, a challenger. And then lastly, you decide what you want to keep or not keep from there. Like just because it says something, I would bet you $5 you put a different day and I don't say the opposite. So like it's only there to give feedback and you do not need to take all of it. AI is not the problem. Using it to avoid thinking is the problem because at the end of the day, your value is not that you can produce fluffy fake words faster. Your value is that you can think and you can make judgment calls and you can make decisions and that you can say something in your own real voice using all the context of your life, like using your brain. Use your brain. That is what it is there for. And that is what makes you so valuable. Now, if you want to take these written communication skills and get better at speaking, you can watch my next video on how to speak like a CEO.