Lately, I felt like I've been hearing a lot of horror stories about the way that people are using AI, not in an AI wrote this program for me, or AI designed the system for me, or even AI wrote this for me, which I think is terrible. Writing is thinking, and if you're not doing the writing, then you're not doing the thinking. I don't care how good you think the first draft is; you should always write the first draft.
But that's not even what I'm talking about. I'm talking about people who use AI for even the most basic of things, like what should I make for dinner? Or what do you think of this work a contractor did for me? You can't come up with your own feedback for the work you're paying somebody to do? Or what question should I ask during this webinar? Think of your own question. Search your brain and come up with a question that you think will actually help you. This is something that I have called AI brain rot. I have a longer piece over at streamlinedsolopreneur.com, but that's not really what I want to talk about today.
I want to talk about my five pillars of AI philosophy. I've been talking a lot about this on LinkedIn and in personal social circles, and I think it has legs and I think it's really important. So that's what we are going to talk about today.
Hey everybody, and welcome to another episode of the Streamlined Solopreneur, the show that helps you automate your business so you can take time off worry-free. I'm your host, Joe Casabona, and here's the problem.
People are asking AI to do too much for them. My friend and I were at an event recently where we talked about, among other things, how too many people are experiencing AI brain rot, where they literally can't do anything without asking AI first. And you might think that this won't happen to you, that you are smarter than that, that everything AI is doing is as a creative partner. And honestly, the term creative partner when it comes to AI is triggering for me.
AI is neither creative nor a partner. It is a computer program. Have you called Excel a partner? Have you called Word a partner? Have you called your browser a partner? Then you should not call AI a partner or an employee, right? You can hire AI as much as you can hire a broom. So maybe AI brain rot isn't happening to you, where it won't happen to you. I don't know you. I know that I've been caught close to letting AI Brain rot take over, and me asking AI to do too many things for me.
But here's, you know, I gave you a few examples at the top of the show, where I see these use cases with AI. Thinking of and creating products is another one. Using it to have a conversation with deceased relatives is another. And look, I know that we all grieve in different ways, but part of the grieving process is acceptance. And my gut is that this will prolong the grieving process or at least prevent you from fully accepting reality.
I miss my grandfather dearly, but I know that it would be extraordinarily unhealthy for me if I created an AI facsimile of him to talk to him. And I spoke with a couple of psychologists about this, who agree with that. I think that is dangerous for our psyche.
And you've probably heard that your brain is a muscle, and it's not. But it is true that if you don't use it, you lose it. And by having AI do more thinking for you than you do for yourself, you will lose the ability to think critically and to do anything well. And I know that we're all different, we have different thresholds, skills, and standards.
But I want to share first. I'm going to share with you three warning signs that you might be experiencing AI brain rot. And then I'm going to give you my five pillars of AI philosophy and how to apply them.
So, the first warning sign is that you can't think of something for more than five minutes before running to an LLM for “advice”. This is the first and biggest. I gave a talk about this in January, I gave a talk at PodFest 2026, and somebody asked me this question. How do you know if you rely on AI too much? And this is exactly what I said. You know, if you can't think of something for more than five minutes before running to ChatGPT or Claude and asking it. If you get some raw material and you throw it into AI and go, what is this? If you get a recording and you just throw it in AI and go like, what are the main points of this? I think that is dangerous. It seems like we're being more productive, right? We're doing things, and we're getting to it faster, but the opposite is happening. We're doing things with no direction. And that is a poison pill for solopreneurs, for our brains, for our systems.
But, there is something worse happening. If I were to put it more extremely, I would say that it's just like any other chemical dependency; you don't feel like you can function without it. And we are already seeing these. People who say, oh, Claude is down. So I guess that's a sign that I should take a break. Let me tell you something, I don't need a sign that I should take a break. I take a break when I think I should take a break. And if Claude goes down or ChatGPT goes down, I keep on working because I can do my job without AI. We need to be able to sit in problems and think about them. Not just for a few minutes, but for 20, 30, 60, maybe more. This is an exercise that we should practice regularly.
I am working on a new opt-in called the Solopreneur Starter Kit. It'll be done by the time this episode comes out. You can get it at streamlined.fm/kit. I have been working on it for days. It is thousands of words, and none of the first draft is done with AI. It is all from me. I'm going to use AI to proofread it and really make sure that I used like the right form of there, there and there. But I didn't say I need an opt-in. Like, generate a PDF of five Solopreneur systems that they should have and how to apply them.
What's the point of that? What value does that add that another person couldn't just do? People want my perspective, not the average of a bunch of people's perspectives. So if you're running to AI at the first sign of friction, couple that with our increasing ability to be bored, and it's having a terrible effect on our collective ability to think critically. So the next time you solve a problem, what article to write, some part of your customer journey, how to talk to your kids about something, I want you to leave your phone in a different room, take out a pen and paper, and spend 20 minutes thinking about it. Or go on a walk and spend 20 minutes thinking about it. You'll build that brain, not muscle, to keep thinking critically. So that was the first warning sign.
The second warning sign is that you're not talking to people who matter. You're letting AI act as a proxy for talking to actual people, customers, stakeholders, business partners, or mastermind members. We need to remember that AI is a massive word calculator. It's really advanced autocomplete. It's the best contextual search we have had. But it is not human. When you ask LLMs something and then treat it as a problem your audience has, you're not serving your audience. You are serving yourself. Instead, again, take out the old pen and paper, spend 20 minutes, and write down five questions you have for your audience, your customers, your stakeholders, whoever you need to talk to. Then, actually email those people those questions. Create a survey or a poll, get real feedback, have actual calls with people. I promise you can use a large language model after that to find primary sources on the problems they say they have. Summarize and process the data you have gathered and show your interesting data points. I did this. I had a bunch of customer calls and coaching calls, and I took them, and I put them into Claude, and I said, I am thinking of doing this product, this workshop. Find me evidence in my customer calls that I should do this. And it said, " You shouldn't do this”. Nobody in any of these customer calls have this problem. So it was back to the drawing board and there it legitimately saved me a bunch of time, but time that I could use to come up with something useful for my audience because I had spent time with my audience. So, that is the second warning sign.
The third and final warning sign is that you have no skin in the game. The final reason, this reason might be the most tangible, tangibly impactful. You have no skin in the game because you didn't actually come up with something, so you couldn't care less about it. There was, you're not a pot committed, right? It's like playing a poker game without real money. Who cares? I'm gonna bet big. I'm gonna go all in on 27. Who cares?
So, when you don't have anything to play for, you don't commit to it. And how does this shake out with AI? Well, let's say you have it come up with a content strategy. You've trained it on your YouTube channel, told it what works and what you want to do, and it comes up with 20 ideas for you to start with. But creating videos is a heavy lift. And sure, it gives you a title and maybe a synopsis, but you're not clear on how to execute this because it's something else's idea. And notice what I said here. Something else's not someone else's. It's something else's idea. I fell into this trap. I thought it was great that it could analyze my content. Everything on my podcast, YouTube channel, blog, and newsletter. I've got over 20 years of content, five talking about systems. But instead of just telling me the insights, it came up with this deep plan that focused too much on a single aspect. But because the work was already “Done”. I decided to give it a try.
But I realized I did not care about most of what it came up with. So what did I do? I went to the old whiteboard. I wrote down four pillars or four categories of videos that I wanted to make that I thought resonated with my audience. And I came up with 10 to 15 ideas for each of those pillars. And those are the videos I'm working on now.
So, you've got the warning signs that you can't go five minutes without thinking about something before running to an LLM, that you're not talking to the people who matter, and that you have no skin in the game.
And if you're worried that you're slipping into the cold embrace of AI brain rot, don't worry, it's not too late. So now I want to give you my AI philosophy and tell you how you can prevent AI brain rot.
I have five pillars to this philosophy. Number one is don't use large language models for thinking work. I'm going to dive deeper into all of these, but number one is don't use large language models for thinking work.
Number two is don't refer to what the LLM does as thinking. I this is a hard habit to break, especially when you say, like what do you think? Now, I say what is your analysis? Or what is the analysis? But it is not thinking. It is running an algorithm.
Number three is don't personify large language models. Notice how I referred to it as it I said something else's idea, not someone else's idea. When we personify the LLM, it's easy to think of it as a living, breathing, thinking person, and it is not. It is a computer program that might not even remember the thing you just said to it. And I know it has memory, “but keeping track of a bunch of context is something that humans do well that large language models do not”. Which brings me to number four.
LLMs are computers and they are best for doing computer things. And number five, remember the cardinal rule of computers, do exactly what you tell them to do. Those are my five pillars. Don't use LLMs for thinking work. Don't refer to what they do as thinking. Don't personify them. They are computers and they are best at doing at doing computer things. And computers do exactly what you tell them to.
So how do we apply these rules to prevent AI brain rot? Use the computer for things computers are good at is number one. That's the first rule. Crunching a lot of data, transcribing meetings Processing written notes. I took a picture of that whiteboard and I said, here are all of my YouTube ideas.
Can you add them to a Notion database with the categories? And it did that spectacularly well. Spectacularly well. Updating apps like Notion and Obsidian with my work, not its work. So good. We are living in an incredible time where we can literally have the full context of all of our meetings, calls and conversations at our fingertips. Then we can have a very smart computer process all of that data for us. This is a great use of AI, but you need to remember to take your own notes to consume the primary source or the primary content. Don't just leave it to AI. Listen to the call, watch the video, read the book, take notes, and then let the large language model supplement that.
I will also use AI for coding. I have complicated feelings about it and there are guardrails you need, but it's very effective at writing code because it's basically speaking its native language here. It's a computer writing for computers. So I've used it to write simple plugins, web apps. I did a whole other episode about two kind of programming based projects that I love. I have my own custom Obsidian theme thanks to it. Really good.
As a former Software engineer, I try to be very clear in my requirements, think about how a human will use the software, and test rigorously. But they've gotten really good at that. So that's rule number one. Use the computer for things computers are good at.
Rule two: Think about the problem you're trying to solve for at least 30 minutes. I know I said 20 earlier. At least 30 minutes away from the computer. When I am thinking through a problem, I like to either go for a walk right in my notebook at my writing desk, where I don't have a computer, or at the front porch or plan at my whiteboard.
Recently I went on, the weather is finally nice here in the Northeast. I went for a walk. There's this big field a couple blocks away from my house. I grabbed a cigar, I took my notebook, I put my phone in do not disturb mode and I walked to this park. I sat on a bench for like an hour and a half. No headphones, no nothing. Just thinking about this problem and it was fantastic. And when I got back, I was energized.
The most important bit is that when you are trying to think through a problem, I say away from your computer because you don't want to get distracted or give up and go to the large language model. You want to have time to sit in the problem and come up with your own solution. It will help you work out your brain so it doesn't atrophy.
But I know what you're thinking. You're thinking, I log ideas on my phone. Joe, this is great that you always have a notebook on you, but I like talking to my phone. And I like talking to my phone too. I did a whole episode on that. And maybe you have the large language model on your phone too. And I have the solution, which is rule number three, remove it from your phone. I know this is a big ask, I know it is. But much like social media, always having it on you means you'll start to use it more as a companion than a program. You have a random thought and decide to ask the LLM. You're struck with inspiration and start using the LLM as a thinking buddy, a hilarious moniker to me because at that point neither party is actually thinking. That doesn't mean you can't use your phone to help you think through stuff.
If I want to take the output from Claude, like analysis of data, for example, with me, I will have it save a file to my Obsidian vaul.
For capturing information or capturing thoughts, I use a completely separate transcription app called Whisper Memos, which I love and I know I've talked about before. That way I can record my thoughts completely and then work with them later. And this includes putting it into Claude and having it pull out your best stuff or what it analyzes is your best stuff. The idea here is to remove the temptation of running to it for everything.
And I'm going to tell a story here cause there is some nuance. We were out in the woods, me and my three kids. There's like a little creek nearby and there's a wooded area and they were having a blast. And my daughter walked through a plant and I'm not an outdoorsy person so I can't identify plants on site, but she got a big rat, like a huge rash and it started to blister. And so I used my iPhone. It's got like that, that camera setting where you can point it at a plant and it tells you what it is. And it's very likely using ChatGPT or some other AI model there. Did I break my rule? I don't have Claude on my phone, I don't have Gemini on my phone. It's built into my phone and I don't think I did because this wasn't a thinking activity, this was a data analysis activity. And I learned that it was a non threatening plant. And my youngest just has really sensitive skin, and sure enough, about an hour later, the blisters went away. So it does take some discretion, but the idea here is that you're removing the temptation of running to it for every little thing.
So, wrapping up, we went over the three warning signs that you might be suffering from AI brain rot. I gave you my AI philosophies, my five pillars of my AI philosophy, and then three rules to prevent AI brain rot. I really hope that you take this into consideration. I get a lot of pushback when I say AI shouldn't write for you. AI shouldn't do creative work for you. If I didn't write it, or if I didn't write it, it wouldn't get written. Maybe it doesn't get written then, because what you're signaling is that I'm not even going to spend time creating a first draft for AI to review. And if it's not worth your time to write, is it worth my time to read?
But that's it for this episode of the Streamlined Solopreneur. If you want my 100% written by me Solopreneur System starter kit, you can find that over at streamlined.fm/kit. It'll give you four systems, tools, automations, and yes, even some skills and prompts to use with large language models.
Thanks so much for listening, and until next time. I hope you find some space in your week.