Stop Outsourcing Your Soul to AI

A beautiful cover image with the text "Cognitive Erosion"

Aug 10, 2025 • 11 min read • 2300 words

It has been more than two and a half years since the release of ChatGPT. The 30^th November, 2022 marked the beginning of a new era. The start of AI. Nowadays, many professions and people in private lives, especially in the information technology sector, use it on a daily basis. And we do not even know where we are on the Gartner hype cycle or if AI will improve following Kurzweil’s “Law of Accelerating Returns”.

The trend is still rising, but at the same time I have noticed a slight but steady degradation of knowledge on a human basis because people delegate more and more creative work to AI. As a result, I have decided to write a little blog post, where I can share my thoughts, trying to discourage the overuse of Artificial Intelligence which would ruin our cognitive capabilities. And although I know that very little people read this, it does not stop me from trying. Otherwise, I might have missed my opportunity in helping humanities future.

Disclaimer: This whole blog post is created from my observations of developments of social changes with special focus on ways of using Artificial Intelligence with my interpretation and is therefore very opinionated. I have not looked at any studies that would support or oppose my opinion.

How people tend to use AI

AI is awesome. I do not want to question this statement. It helped and still helps me improve my workflow, gathering new knowledge, diving into new waters and exploring new technologies. For people who are not interested in how AI works, such intelligent tools seem like magic. As a result, I like to see their eyes sparkle and their mouths left open when I explain to them that everything AI is based on is probability. What word is most likely to be the next one based on properties and emotions in a highly dimensional virtual space. But that is another topic.

What I would like to focus on is how people have come to use AI, especially over the past months - if not years. Since our brains are lazy by default, we tend to outsource more work to AI the less focused we are on a specific task. In other words, if you do not put most of your commitment into your current project, you are likely on the verge of just prompting an AI with it. Another potential reason for offshoring work to the AI world is ignorance of the underlying technologies, which makes it difficult for you to come up with your own solutions.

I am by no means a scientist, but I think that watching AI doing your work - because due to your incompetence the outcome you produce is not aligning with your vision - gives you so much dopamine that your brain wants to do it more often. Moreover, the human being as a social figure always aims to be connected to somebody and engaging with AI satisfies those needs. This behaviour is typical for our brains and also happens when watching social media or consuming explicit content. It is deeply marked in our cerebral cortex and the whole gaming industry and entertainment sectors use it recklessly to optimise for money.

I noticed this trend myself, as you can clearly see from some of my posts where I admittedly used AI to help me write them. But I think this is not a shame. Every content creator who uses AI should experience first-hand the pitfalls of creating content directly for the viewer or reader, in order to realise how little they subsequently know about the topic. This will enable them to learn from their mistake and be aware of how easily laziness can lead you down such a despised path. Only the excellent among us strove for hand-crafted content all along.

Negative side effects

I pointed out the main drawbacks of using AI too extensively that I observed. I will now give you two smaller reasons I have noticed in myself for why you should not use AI for expressive work.

Communication decline

While watching a promotion video from “The Browser Company™” (see the last link in the resource section below) I noticed that visions can be very far apart nowadays. I see this trend not only in the Dia browser, but in many different aspects of technology and parts of our lives. One side of the trend wants to integrate AI everywhere possible, widening our dependence on it. One example is using AI to compose and write messages to our closest friends and colleagues because it makes us more efficient. I believe that it just makes us stupid in the long run.

If people do not train their brains for coming up with words for communication, building useful sentences and meaning that can be understood by other human beings, our very foundation of interacting with each other is broken and we would be back at the level of Homo habilis - the last species that likely did not use fully developed verbal language. See, I used AI to help me research this very bit of information, but chose my words consciously to express myself. This way, I not only learned something new, but can now also remember it. A capability you will never have when outsourcing writing to AI.

Loneliness

Being alone on a mental level is in my opinion one of the most underrated modern dangers of society. The ironic part about it is, that if everyone is alone, we all have it in common, making us not alone with each one’s loneliness. But my thoughts are wandering in other oceans again.

AI strongly encourages us to be alone. It ~~acts~~ was intentionally trained to act as a perfect friend, colleague or whatever you wanna call it. Recently I noticed that ChatGPT’s responses explicitly include my name which feels very weird on multiple layers if you ask me. One could argue that this is private artificial doxxing, which of course is a oxymoron in itself as doxxing by definition is public. Nevertheless, the more you use AI, the more often you want to use AI. Particularly after the introduction of AI voice modes this trend kept rising like there is no tomorrow. Congratulations ClosedAI, you just reinvented addiction.

In worst case scenarios, this can leads people to fear talking with other people. Abandoning friendships. Withdrawal from society. Such habits are serious problems and should best be examined by a doctor or psychologist.

Proposal of AI usage layer model

Proposing a global solution for all people of such a problem is impossible. Since each individual has their own experiences with AI, their own thoughts and feelings, optimists or pessimists, I cannot speak for everyone. In fact, I think I can only speak for myself, learn and share my own ways of handling and limiting my usage of AI and hope that those experiences can help you as well.

My ideal experiences with AI - the optimal golden way between the overuse of AI alongside cognitive erosion and letting the hype pass besides you without taking any advantage of it - is using AI iff you and only you gain any benefits. In other words, control your usage and limit it to an extreme, where only ever you should come into contact with the content the AI produces. Such a limitation has many consequences, here are some examples to better demonstrate and emphasise what I mean:

Use AI for personal learning sessions: If you initially want to dive into a new, completely unknown topic, AI can give you general overviews and show you connections to other parts of knowledge and real-world uses. While this does not necessarily mean that you really learn faster or more efficiently, it helps to take the step forward and get the ball rolling.
Use AI for stupid and repetitive workloads and ask for help if you are certain that the problem or task has already been solved thousands of times by others. Current LLMs are trained based on existing data which means they are meant to be good at solving solved problems, not reinventing the wheel, but suggesting to use the existing round instrument.

Do not use AI to express yourself to other people. Think about what meaning you want to convey, what knowledge you want to share and pack your thoughts into your own words. This helps you immensely to harden your own thoughts and remember what you learned and feel much better.
Do not use AI to generate complex and creative work. Nowadays, there are hundreds - probably thousands - of tools, which help you create art, music, code, designs, plans, etc. with the help of or completely with AI. While this makes it seem that you can be more productive, it just makes you more ignorant the more often you fall back to those methods instead of doing what you love to do yourself. It is completely fine to say that small packages of the work can be outsourced to AI, mostly repetitive tasks that just take more time doing manually. Creative and original work, however, should remain human and made with love. If you do not like what you do at all, this is a signal that you should probably stop doing it at all.
Do not fall for all the shiny AI stuff out there. Tools like Lovable, GitHub Copilot, Cursor, ChatGPT Voice and new on the board: GPT-5, just to name a few, do not want to actually help you. They want either directly your data, or own your workflow so you cannot escape once you got used to it. Unfortunately, people get used to AI fascinatingly fast.

In short, what I mean is that AI should be used for brainstorming and helping each individual for creating an end product - not to be confused with AI creating the end product -, but not for communicating with other people and the final results themselves (eg image creation, writing blogs, posts on social media, communicating with chats, motivation speeches, essays, etc.). In my opinion, the human brain can only evolve and learn from its own mistakes and experience. And you can never learn from your mistakes when it was actually the AI generating them.

AI has a place in ideation, but expressive work should remain human.

The Paradigm

I recently discovered a software design pattern called the “Onion Architecture”. To keep the explanation short, the main key principle of this architecture is its division into layers, where each layer is only connected to its two neighbours (siblings excluded), the inner and outer layer.

To transfer this paradigm into the current context, you can imagine the AI as a tool available to you, which is unconnected to all other parts by being only surrounded by your brain, and only your brain can access, manipulate, create, and modify the outer layers - in this example, writing blogs, creating content, communicating with people via chats or speeches, or expressing yourself with art or design.

Here you can see an image that represents my idea of a limited, but ideal usage of AI:

Cognitive Erosion Paradigm: Layered, green circles with "AI" in the centre, surrounded by the "Brain", which is surrounded by multiple circular segments labelled with "Design", "Blog", "Chats", "Art", "Content", and "Speech" clockwise.

If such an architectural approach and the paradigm in this context are optimal or just opinionated can be criticised in the comments below.

Conclusion

As I said in the beginning, this blog post is very opinionated and aims to just share my experience with everyone so you can also learn from my mistakes and optimise your own usage of AI and have it better under your control. In the end you can do whatever you want, but be aware of the degradation of your own brain when you also fall into the pitfall of overuse.

Also one thing to note: It took me several months to write this blog post, refine it and start from zero again. No blog is perfect as you might have noticed with this one. There are probably some grammatical error here and there, some sentences that actually do not make sense, but this makes humans human. Moreover, there are very likely many strong opinions in this blog where you would totally argue against and completely disagree with me (I’m eager to see your comments expressing your outrage.), however this makes me me. My brain, my thoughts, my opinion.

“errare humanum est”

Resources with better evidence

If you are interested in articles which are based on more scientific studies or approaches, feel free to read some or all of those. They are in no particular order, so just start in the middle and keep going upward or downward. 😊

Happy reading!