I banned AI to myself for 7 days to get my mind back

Jul 7, 2026

17 mins read

For the first time in modern times, we’re seeing human intelligence being treated as a replaceable thing. No need to learn skills anymore or hire people who have learned them. No need to research things on your own or think about them. AI replaces your brain and does all of that for you at a much larger scale in much less time. Just buy more tokens from AI companies instead.

This is the stupidest idea ever. Not just a stupid idea, probably the stupidest idea. It is in the delusion of achieving being smarter than ever, but in reality what is really happening is the replacing of real human intelligence with a program that guesses the most probable words rather than actually thinking.

AI of today is just a machine that tries to come up with the statistically most likely continuation of a given text. It’s not something with real intelligence. In fact, I am not even sure if it is correct to use the term “Artificial Intelligence” for this current technology. But AI companies don’t want us to realize what “AI” actually is. They just want us all to think that:

“FINALLY AI IS HERE AND WE ARE NOW ULTRA INTELLIGENT! WE WILL BECOME A UTOPIA, FIX ALL PROBLEMS AND BE CRAZY RICH WHILE LIVING FOREVER!!!”

They want us to think like this this because they want to sell their product and make profits. They keep pushing the “AI” to us to use for everything from programming, writing, researching and so on. They also create the fear of falling behind and losing our jobs if we don’t use more and more “AI” every day.

I think this constant chase of using them more and more for everything is harming us. Chronic use of AI for everything results in distracted minds that struggle to think deeply about anything. Making us stupider and dependent instead of smarter. This is what I experienced to happen to me after a long time of heavily using it every day. I decided to quit and evaluate what’s going on and that’s what I’ll talk about in this post.

My experience using AI

I have been doing programming since 2020. At the time, I started with Arduino projects using the Arduino IDE. It didn’t even have intellisense. And when I found out about VSCode in 2021, it felt like such an upgrade from the Arduino IDE. Back then I used to research and code everything from scratch. This is how I operated in both learning programming or researching other things (like everyone else).

And then “AI” came out and they promised us so much productivity benefits. Who doesn’t want to be 10x productive? And it’s easy, just ask a question to ChatGPT or Gemini and let it stream to you it’s response. It searches the internet in seconds. Actually it kind of is the internet distilled into a program, right? So we better keep using them for best productivity or we will fall behind and lose our jobs.

For the past few years, I’ve been using AI chats and AI development tools to be more productive. I kept using increasingly more AI as time went by, both for coding and researching things I am interested in.

Half my open browser tabs became Claude or Gemini. Whenever I got an idea that I needed to think deeply about, I would always feel the urge to go ask an “AI” about it and “research” it that way. Then it would turn into a constant back and forth with the “AI” a some time to get all I needed.

I realized that I wasn’t really focusing and thinking deeply about things by myself much anymore. Whenever I wanted to sit and think about something, I would start chatting with an AI. And if I wanted just think by myself instead of using the “convenience” of the LLM chat, I couldn’t do it for long. I would either get distracted into another topic or come back to the AI eventually. I would struggle a lot when I tried to focus on a topic on my own. I was losing the ability to focus and think deeply on a topic for long durations by myself.

Also after hours of using “AI”, I would end up feeling pretty tired. I would feel overwhelmed with the amount of information that came from that AI chat and feel lost. I think my memory was getting overwhelmed with the size of the unstructured context that it tries to hold.

It was a not a good situation and affected my working a lot. At the times I used AI agents like Cursor for coding, I would struggle to get myself to write even simple functions. For example, a function that lists all the hours between waking up time and sleeping time felt hard to make. I was also unable to focus on writing long blog posts or other long writings and every time I drafted something I would get the idea to ask an LLM how it looks. After I get a response, I would make adjustments and ask again. And I would end up in a loop of doing that for a long time instead of focusing on writing the thing.

AI was making me lazy and dependent. It was making doing anything on my own painful because that required more focus and meticulous effort than just using AI. I wasn’t only losing the ability to focus on things on my own but I was also losing the desire to focus. Using AI was always the much easier action to take. Though not necessarily the most accurate or actually productive.

I realized that I would keep going with the “AI” chat for hours not just because it was helping me get things done faster. It was because it was easier to use it and focus on it then doing things on my own. It became a habit. Whenever I’d think of researching something or writing code, my mind would immediately think of the AI tools.

I realized that I wasn’t being really productive and my mind wasn’t working right. It wasn’t able to do anything complex on it’s own. It was constantly seeking using “AI”. I tried to use less AI or quit but it didn’t work because every time I tried to something without AI my mind would scream and make me think it’s a waste of time.

So because of this reason, I decided to use my strongest weapon against this situation. I used my discipline contract method to ban myself using AI completely for 7 days. So that I could see and remember what it is like not using AI. So that I could see how much my mind was affected from using it.

How was quitting AI after years of heavy use

It was very painful. I kept constantly getting the urge of using AI and my mind kept screaming to me “this is slow!” every time I tried to do something on my own. But I knew that this slowness feel wasn’t coming from my logic, it was coming from my state of being used to using AI.

It was so hard to focus on a topic and research it or write code. I would feel very bored. I would open the codebase, stare at the IDE for hours and not get anything done. My mind would keep wandering into 3-4 different topics at the same time: a life situation, a game that I played, a song playing in the background of my mind…

But it got better towards the end of the first week. As I pushed through and did lighter tasks in the codebase, I could increase the scale and complexity of the tasks gradually. My focus was improving as I suffered through the distracted horrible state and let my brain calm down.

Improvements happened as a slow progress over days. And at the 7th day, I was significantly better but I still didn’t feel like my past self before the time I got so used to using “AI”. But it’s normal, 7 days isn’t much for recovery. After 7th day I stopped evaluating each day but I kept getting better. Right now as I post this, it has been 23 days and I only used AI a few times for simple things like checking grammar issues. No chatting back and forth.

I understood that it was true, my mind was really affected from all this constant “AI” (ab)use. I was finally relieved that I could get out of that state and get my brain working right again after years.

AI Addiction

I realized that what happened to me after years of daily AI use for everything is probably a type of addiction.

See, the current “AI” chat tools like ChatGPT or Claude Code are all based on the same thing:

You send a prompt explaining what’s in your mind
LLM processes it and streams a response for you
You read and evaluate the response
You get the idea of the next prompt and type it in
It streams a new response to you

And this loops for as long as you like or run out of tokens.

What this is, is an endless feedback loop. A feedback loop is when you get into a loop of taking an action and getting a reaction to it. In the case of AI chat, you do the action of sending a message and LLM reacts to it with streaming you a response. You read it, think about it, send another message and it continues on.

What this does to your brain is that it trains it to keep wanting to do actions in this feedback loop. Since the LLM response messages feel interesting to you, your brain releases dopamine from sending messages. Over time, your brain learns to associate messaging (the “action”) with the pleasant feeling that follows the model’s reply (the “reaction”), making you more likely to repeat the cycle.

It’s like slot machines

Slot machines get people hooked the same way. You pull the lever and expect the spinning slots to match so you can earn money. It is actually a statistical game and you can calculate your chances of winning for how many lever pulls you need to do (assuming it’s not rigged). And you can make a decision based on your calculations for if you want to take the risk.

But that is not what happens. Instead, you just pull the lever without thinking. It spins the slots with a cool animation. Everything looks so colorful and luxurious. Spin ends, you lose. You try again. You win a small prize. That makes you keep believing you can win. You then continue pulling until you lose too much and understand that you are fooled.

Why did you pull the lever in the first place? It’s not because it was the logical choice to pull the lever. It was because the thrill of winning made you so hyped up so that your logical reasoning got overrun by the thrill. This led you into the feedback loop of getting dopamine to pull the lever each turn.

It is the same feedback loop that makes you seek more of it due to the association of action and reaction. Your emotional system feels good doing it and your logic is convinced of the possibility of winning, so you keep doing it until you lose all your money or your logic finally understands that you’re being an idiot.

Brain’s tendency for choosing easier and quicker rewards

Our brains are inclined to take the quickest path to reward. And the more it feels hardship, the more it is inclined to sway to the easiest path. Paths that give reward in more time and effort deplete dopamine more and feel harder to follow. Especially when easier paths to getting reward are present, harder paths feel even more harder.

Thinking deeply, reading sources, searching the web yourself become the harder path when there is an AI chat. With AI, no need to structure your thoughts and process them yourself, no need to spend hours, just ask ChatGPT. It distracts you away from the harder things. Just like how social media or video games does. On the case of AI, you are convinced it gets you what you want faster therefore it is more productive. So your brain’s both emotions and logic makes you tend to choose using that easier and quicker option.

Also, constantly using something makes it even an easier path to choose for our brains. The brain gets familiar with doing it so many times that the easy option becomes even easier.

Real thinking

Archimedes yelled “Eureka!” when he realized that his body would displace the exact same volume of water as his body when he entered his bath tub. He understood that he can measure the volume of irregularly shaped objects with the method of submerging them in water. Galileo proved that the Earth wasn’t the center of the universe because he understood the movements of the celestial bodies, thought and researched about it for years. Every science achievement, every engineering achievement came from people thinking deeply about and researching a subject for a really long time.

Any profession that deals with complex cases of problem solving requires deep thinking for long durations. Whether it is science, programming, graphical design, law, medicine, engineering, economy, architecture, business, strategy… We learn complex concepts, we develop mental models, we think deeply each time we face an issue to solve or a project to make.

AI can’t think

What the thing that is called “AI” does is not thinking. All it does is just statistically guessing the next words of a text based on their training data.

This method works for some appliances and it is useful as long as you understand what it really is. For example, it works for things like summarizing large volumes of data. Though even at this, they aren’t always accurate at bringing facts as they are. I have personally seen it reverse the facts and I wouldn’t be able to catch if I didn’t know the truth myself.

We all have seen how they fail so badly when they are asked things that require logic that they don’t have an example data about, like how many "r" letters in "strawberry" or what happens if you let go one end of a pen that you were holding both ends of. They tell you ridiculous answers with 100% confidence. This is because there isn’t a real intelligence processing the meaning of the words with logic. They just make statistical guesses based on whatever data they have.

When these errors happen, AI developers fix them with adding the relevant data into the training of the next version of the LLM. Like adding mathematical formulations and calculation examples, so that it makes the LLM more accurate at responding correctly.

The actual thinking is done by the people who create the data. They find the solution to the problem and put the data in that LLM. LLMs just processes the data in a mathematical way to fetch you the relevant data to your input. If they had real logical reasoning capabilities, they would never make mistakes on things we humans find trivial.

Real work and information comes from the people who created the information and put it on the internet. AI developers take all this information and shove it into a computer program to process the statistical relations of the words. This creates the thing called LLM. Which are basically robots that determine their actions with rolling dice.

LLMs can’t understand complex topics because they don’t have a real mind to understand them. They don’t develop mental models because they don’t have real logical reasoning capabilities. They can’t discover things that they don’t have data about. They can’t think for you.

We can’t replace human thinking with AI

Real thinking is what really creates new information, new thoughts, solutions to problems, brilliant ideas, breakthroughs in our understanding of things. And it is done through the real reasoning by a human mind. Can’t be done by a word guessing program.

LLM technology only does probabilistic predictions. It can’t have the qualities of ability “to know” or “to understand” at all, because those are the qualities of only a real mind.

I acknowledge that the LLM method is useful when utilized the right way. It works great for things like quick information retrieval or summarizing things. But it can’t replace the real creative and problem solving intelligence of the human mind. They should be used for what they are good for, while also being careful for their inaccuracies.

Also, a generic mathematical average calculation algorithm should not replace our unique ways of thinking. All of us have unique brains developed differently thanks to our our own unique experiences and physiologies. That uniqueness is what makes us sometimes see and discover things others can’t. Using LLMs to think instead of ourselves takes that uniqueness away too.

Trying to replace our thinking with “AI” is also harmful for our brains. Constant exposure to the immediate feedback loop of the AI tools trains us to give up on our own thinking and destroys our ability to focus for long periods of time. It becomes a trap which we end up dumber instead of smarter.

We have to be very careful

Every new technology comes with very exciting benefits but also produces new pitfalls. And when we are so excited with the benefits, we don’t to see these pitfalls until we get hurt from them and excitement withers away.

We found fire, which you can use to warm yourself and cook your food, but also it can burn down forests.
We found cars, a great tool to get from a place to another, but it is also possible to lose control and crash with them.
We found antibiotics, which allowed us to treat infections, but overuse of them cause development of antibiotic resistant bacteria species.
We found plastics, which allowed us a massive opportunity of producing cheap and durable tools and parts for our machines, but they also pollute our oceans and our food like nothing else can.
We found nuclear energy, which provided massive amounts of energy, but we can also cause horrible radiation leaks or explode each other to extinction with it.

I can keep going but I think this is enough to make my point.

“Artificial Intelligence” technology is no exception. And unlike all the other technological developments before, which had dangers to us through our environments, this so-called “AI” technology has dangers on our thinking processes.

“AI” tools get us hooked to use them more and more. They change our behavior in the way that they keep making us use them more. Our minds get affected by the convenience and the behavioral training we get with the feedback loop of using them and that is shaping our actions for when we want to work and think.

I am not saying that we should completely get rid of this technology. I am just saying they need to be used right. They need to be used for what they are good for. They shouldn’t be used in way that they hurt our minds and make us incapable of our own independent deep thinking.

We must be very careful. Those who don’t, get hurt from every new technology. Those who do, find ways to use the technology carefully without hurting themselves. Today it is LLMs, tomorrow it may be something else. Every time a new tech comes out, try to be the ones who sees the pitfalls too. And in order to do that, we need to understand what they really are.