Question-01¶

Critical Thinking and Creativity¶

As the AI Era unfolds (is it, actually?), in June 2026, on the Internet, we can meet a specific type of publication expressing concerns about whether the usage of AI will damage or even kill critical thinking and creativity.

However, before going into the subject of the interaction between humans and this new tool (AI), it seems to be necessary to take a wider look at the problem: what was the situation with critical thinking and creativity before AI? Before smartphones? Before the Internet? Before computers? While it is extremely difficult to generalize data for any conclusions on that, we can still have this feeling that the situation does not change dramatically with the arrival of any new tools: critical thinking and creativity strangely remain the feature presented in a very low percentage of the population. So, the question of how AI will affect critical thinking and creativity dissolves, as there is simply nothing to be affected in the first place.

Instead, we obtain the question "how can AI affect those adults who have the trait of critical thinking or some degree of creativity, and children who have potential for that?"

The answer to that: AI will not do anything in that regard. The only question is what those people will do to themselves with these tools in their hands. We already discussed the nuanced usage of this tool in the "Testing" section of this project, and based on technology features, have concluded:

"That is not AI "who" "creates" the "answer" - that is fully Human, AI only populates it with details and wording. Initial and general answers are always generalization of other Human consensus in the simplest possible form."

It absolutely depends on the intent of the person using AI (LLM) if it is used for research (where you ask extra difficult questions, doubt the output, check the cited sources, combine AI output with different types of information you can access and explore) or for thinking outsourcing (where you ask to provide you the answer without bare understanding of this answer an no clue of it's meaning and bias). In that sense, AI remains an amplifier tool in terms of information access (when you ask the right questions) and creativity (where you just stop using AI completely every time you feel its output does not reflect your vision and inner world).

While AI's owners put at the bottom of their UI the note "... is AI and can make mistakes, including about people" to protect themselves legally, to protect critical thinking, we could replace this with "This is just an opinion. There are others. There are missing details."

About children: we (adults) do not usually leave our children alone; thus, those parents and teachers who have the traits of critical thinking and creativity will share them with their children, and for those of them open to that knowledge, that will work perfectly as it always did. This includes, nowadays, an explanation of what AI actually provides ("answer" not "Truth") and how to use it.

Generative AI and Social Pressure¶

Under this project, we have already established that AI is a Multi-Window Reality. While in the Critical Thinking and Creativity section, we were mostly talking about chatbot variants of use and "free of social pressure" use of those chatbots for research and creative tasks, we also need to pay some attention to the use of generative AI ("AI agents") for coding and mundane content generation (statistical reports, corporate diagrams or charts, summaries for some day-by-day information flows, etc.) These usage scenarios potentially affect the skills and well-being of professionals using AI.

Unlike chatbots, AI agents are not "conversational" but "prompt-to-final-artifact" tools. In the case of coding (developers), this creates several potential risks:

• Outsourcing vs. Augmentation: A developer who uses AI to handle the "boring" 70% of a task to spend more time on the complex 30% of logic/security/scalability is using it as a tool. A developer (or PM, or anybody) who uses AI to complete the task because they don't understand how to do it themselves is "outsourcing their brain." If the AI produces a complex solution and the human doesn't understand the underlying architecture, they cannot "debug" it when it breaks under unforeseen load.
• Supervisory Overload and Atrophy: A developer today spends more time reading AI-generated code than writing their own. This requires a specific skill: Code Literacy. You must be able to parse logic flows and identify "code smells" without having performed the cognitive labor of building them yourself. A human who delegates coding itself to an agent risks losing the ability to "debug" the agent’s logic. By "automating" the generation of code and reports, we have not eliminated labor; we have simply shifted it to the most mentally taxing and unrewarding part of the process: Post-hoc Verification.
• Harmful Centralization and Cognitive Load: In a team, a technical writer questions the logic of an engineer; a QA catches a bug the developer missed (inter-subjective verification). When you centralize the role of engineer, architect, QA, and writer into one person using AI, you aren't creating a "super-efficient entity." You are creating a single point of failure that is blind to its own hallucinations. At the same time, cognitive load for the "person-project" increases to a potentially unsustainable level.
• Social Pressure and Burnout: The developers "with AI agents in their hands" are expected and demanded to work much faster and provide more results, while they still need time to learn and think. The social (management) demand for "more results, faster" forces individuals to bypass both learning and thinking to hit their deadlines. If social pressure forces a developer or author to outsource that struggle to an AI to meet a quota, they are effectively trading their future expertise for a short-term performance boost. Besides the degradation of expertise, that also potentially leads to faster and more severe burnout, as you work fast and do not extract any cognitive satisfaction out of the process.

The risks listed above apply not only to developers, but, in the same manner, to mundane content (statistical reports, corporate diagrams or charts, summaries for some day-by-day information flows, etc.) creators.

Thus, we come to the conclusion that using AI agents is potentially much more harmful for the people using them or being forced to use them than using LLM chatbots for general purposes. This potential harm becomes real in the social context: when people (not AI) misunderstand the productivity, progress, and well-being of other people and their own business.

Current Conclusions and Open Questions¶

Current Conclusions:

1. AI (LLM chatbots) is not a threat to critical thinking for 95% of population as those people do not have any critical thinking. AI obviously does not by itself contribute to the birth of critical thinking, but other sources of information (schools, books etc. do not do that either).
2. For the remaining cases, how AI affects a user, fully depends on human intent of that user. AI can potentially provide a huge contribution into critical thinking and be (by intent) excluded from specifically personal creative expressions.
3. Adults traits of critical thinking and creativity are free to choose to educate children on nuances of AI usage and on differences between "answers" and the "truth."
4. AI agents are potentially more harmful for human skills and well-being than LLM chatbots. This potential for harm is actualized and can be neutralized by social balancing and boundaries.

Open Questions:

1. What social institutions are capable or responsible for fighting for social balancing and boundaries mentioned above? Trade unions? Are they even capable of affecting modern corporations?