AI-Assisted Cognition Endangers Human Development

To understand what AI-assisted cognition is, we first need to understand what cognition is.

“Cognition is the mental processes that relate to knowledge. They involve psychological activities that acquire, store, retrieve, transform, or apply information. Cognition is a pervasive part of mental life, helping individuals understand and interact with the world.” Question: Wikipedia

Cognition can be aided by external static information or external cognition.
For example, most people would classify reading a book and discussing a topic with another human in the category of external static information, because humans themselves think and process information in the category of external cognition.

But where does the discussion fit in with AI? They are able to process information that can result in original solutions^{But they are still stable and cannot learn at present.}

In early 2026, the United States prepares to invade Greenland and, therefore, the European Union^{. Just a few months before that it was completely unimaginable that the United States would even think of threatening an invasion of Greenland. Since AI base models are stuck in the past, they easily do not consider these events as real and often label them as “hypothetical”, “fake news” or “impossible”. It also affects newer models like Gemini 3 Pro, GLM-5 or GPT-5.3-codecs.}

Since most new LLMs are post-trained on relatively old base models, even when post-trained on new events, they do not fully utilize this information in their cognition and are still biased toward stable patterns of the base model’s hidden states.^{. They basically think something different than what they say.}

so you can see the problem ^{Already here: If many people use AI to discuss, write, autocomplete, and brainstorm, but AI cognition does not reflect new events and cultural changes, such as changes in relations between the United States and the European Union, new geopolitical realities, and attitudes of the EU population toward the United States, then people will gravitate toward these old patterns and ideas. Cultural change will have to be sustained and sustained indefinitely to hold up against the static cognitive biases of AI.}

Human knowledge and thought, and thus human development, are highly dependent on a dynamic dialectical substrate^{.
Understanding the dynamic dialectical substrate will help understand how AI-assisted cognition can threaten human evolution and how to use AI-assisted cognition without endangering human evolution.}

Dynamic dialectic substrate is the sum of all local and global dialectics ^{Processes and findings. It is the fundamental foundation upon which all of humanity is built, and it is the root of all ideas, concepts, thoughts and solutions that humans use.}

Dynamic dialectic substrate creates new concepts through a process of qualitative merging of existing concepts, which may occur within an individual, a group of people, or even globally.

The above image is a narrow slice of the dialectic process present in the dynamic dialectic substrate. You can see how concepts merge and evolve into higher and higher concepts. In this example the following dialectical process emerges:

first stage: :

• The result of “cold is painful” and “fire is hot” “Fire takes away the cold and pain”
• The result of “vital water extinguishes the fire” and “rain is falling” “Heavy rain put out the fire”
• “Rain is falling water” and “The roof of the hut is” is the result of “Hut-like shelter to protect from rain”

Step 2: :

• “Fire drives away cold and pain” and “Rainy rain extinguishes fire” result in “Rain extinguishes fire and hence causes cold and pain.”
• “The fire is extinguished by the heavy rain” and “The hut is sheltered from the rain” results in “Inside the hut, the fire is saved by the rain”

step 3: :

• “Rain extinguishes fire and hence causes cold-pain” and “Inside the hut, fire is saved by rain” Result “The hut protects the fire and hence protects from cold and pain.”

Because LLMs prefer or lean toward certain patterns and concepts (known as inductive bias), even after training, they have reduced cognitive thresholds when used as a tool for cognition at the population level. This is especially true if only a few AI models are used, or if multiple AI models share only a few base models. This will destroy the diversity of ideas, concepts and solutions, thereby slowing down human development.

You can think of it as a world in which a significant portion of the population is talking to the same five people to discuss problems, the world, relationships, and basically anything. It’s hard to say how much of an impact these five people will have on humanity, even if they try their best to remain as neutral and open as possible. The thinking of humans who talk to these five people is still massively changed, and this becomes a significant problem at the population level.

It’s entirely possible that we’ve already missed great scientific discoveries or cultural shifts due to AI-skew or simply not noticing.

I tried to visualize this problem in the following image, which shows how the range of higher level concepts is skewed in the direction preferred by the base model:

Coming back to the example of the United States invasion of Greenland: it is clear that humans using AI to brainstorm the geopolitical future of the European Union, the United States, and Greenland will encounter patterns skewed toward the “worldview” of the base model. This bias may also prevent many people in the EU from considering the possibility of moving away from foreign services or software. Such a change could have massive consequences, especially since the EU relies heavily on USA services and software that could be shut down at any time. If this AI-skew affects even single individuals from specific groups such as politicians, CEOs, managers or scientists the impact may already be significant due to their decision-making power.

Because it is incredibly expensive to train the base model and carry flexible biases, those who do not have access to a GPU cluster have to accept that these issues exist. To avoid problems like AI-skew and going unnoticed, one should focus on using specific strategies to mitigate them.

Speaking and discussing with other human beings is clearly the most effective way to reduce these problems. It would also be wise to mention that if you already have a good idea of ​​a solution through AI-assisted cognition, you have to be careful not to push other humans in your direction. Try not to use questions or prompts that would lead other humans to the solution or idea you did through AI-assisted cognition, unless the other person is exploring a cognitive path you haven’t discovered yet.

With regard to solutions that directly involve the use of AI, our range of options is quite limited, and there is no solution yet that can fully or partially solve this problem at a population scale. Here are options that at least broaden the range of concepts and ideas one can get from an LLM, while unfortunately not mitigating the main problem:

• Use search engines to find relevant sources of information or let AI search through you Web Search And prevent it from giving you any solutions or ideas directly.
• Use different types of AI with different base models
• Explore different “AI personalities” that simulate different attitudes and thinking styles such as: “You are Einstein”, “You are on Drug X”, “You are a neurotic but dignified sea otter”

Even though we have hints and even some evidence that AI-assisted cognition may threaten human evolution, the extent and depth are still unknown and unclear. More outcome-focused research is needed to understand the significance. Since we don’t have another humanity to A/B test, there will always be a lot of uncertainty and speculation on this topic, because if one wants to participate with other humans or their creations, one cannot separate one’s cognitions from the influence of population-level AI-assisted cognitive biases, which must already be influenced by AI-skew if it is to have any significant impact.

It’s not entirely clear to me how we would recognize the effects of AI-skew and ignorant denialism at a population level. We can’t know what innovations, discoveries, and cultural changes we’re missing out on because of it. Although I’m sure there will be statistics that will offer small clues into the all-consuming cataclysmic stories, as I may here a little for logic and attention to be consistent with our shared attention economy, it is probably, like everything else, not that simple.

It is not easy to imagine a solution to all this, but I, for my part, will definitely try to adopt more “cognition hygiene”… Besides, for me it is much more fun to talk about ideas and concepts with humans than with AI.

I’ve noticed slow awareness of this incredibly important topic and I hope I can speed it up a bit with this article and by giving people a framework to understand and speak about it. If people don’t have words about something, it’s difficult to think and speak about it. It will be interesting to see how this topic evolves.

The topic of AI-skew and AI-assisted cognition is full of unknowns and it would be good to talk to people about it. I hope this article can be a starting point for that. If you want to share your thoughts, or are interested in a conversation about that, you can mail me [email protected]