Thinking and Producing with Artificial Intelligence (Article 06)
The transition from giving commands to establishing communication and the emergence of an interactive thinking model
Aydın Tiryaki and ChatGPT AI (April 25, 2026)
Introduction
Most users who encounter artificial intelligence for the first time tend to treat it as a tool. They give a command and expect a result. This approach is a natural extension of the traditional software mindset we have been accustomed to for years. However, it does not take long to realize a fundamental difference: artificial intelligence does not behave like a classical tool.
You provide a command, and it generates a response, but that response is rarely final. It is usually a starting point—something to be discussed, refined, and improved. Once this distinction becomes clear, the way of using AI begins to change. Giving commands gradually gives way to building dialogue.
Dialogue Is Not Only Written
When we think of interacting with artificial intelligence, written communication often comes to mind first. Yet today, this interaction is no longer limited to text. Many AI systems support voice input, and some can also generate spoken responses. This creates a more natural, conversational experience between the user and the system.
Although these systems are not yet perfect, voice interaction has already become a valuable alternative, especially in situations that require speed and fluidity. At the current stage, AI cannot directly interpret tone, emotion, or subtle vocal nuances. However, it clearly derives meaning from the structure of sentences, the choice of words, and the overall expression, and shapes its responses accordingly.
From Commands to Dialogue
In traditional systems, giving the correct command leads to the correct result. The system does not argue, guide, or propose alternatives. Artificial intelligence, however, behaves differently. It interprets your input, fills in gaps if necessary, expands or narrows the scope, and sometimes approaches the subject from entirely unexpected angles.
This flexibility is a powerful advantage when used properly. But if one insists on interacting through commands alone, this advantage quickly turns into a limitation, because the process remains one-sided. True efficiency emerges when this one-way interaction evolves into a mutual exchange.
What Is Dialogue?
Establishing dialogue with artificial intelligence means moving beyond a one-sided chain of commands and entering a shared thinking process. In this process, the user is no longer just someone who asks questions, but also someone who evaluates responses, challenges them when necessary, guides the direction, and actively shapes the outcome.
Responses are not simply accepted as they are; they are examined, questioned, and reconstructed.
The First Response Is Not the Final Response
One of the key realizations in working with AI is understanding that the first response is rarely the final one. The system produces an initial output based on the current context, but that output can almost always be improved.
At this stage, the user becomes part of the process by asking follow-up questions, proposing alternative approaches, or directly challenging the response. Through this interaction, the result evolves into something more refined, more meaningful, and ultimately more useful.
An Experience: Change in Addressing Style and What Followed
A particularly striking example of this process occurred during a dialogue I had with Gemini. While dealing with issues in the voice input system, I suggested a dual-language input structure to address the difficulty of recognizing technical terms, which are often used in English.
In response, the system noted that graduates of certain universities—specifically METU, Boğaziçi, and Bilkent—tend to use such a mixed language structure more frequently.
I then mentioned that I was a graduate of METU.
This small piece of information immediately changed the system’s approach, and it began addressing me as “Aydın hocam.” This form of address is quite natural and widely used within METU culture. What was remarkable was not the phrase itself, but the system’s ability to capture that cultural context and reflect it in the interaction.
For a while, this form of address persisted and was even incorporated into the system’s personalized settings. However, in different situations—especially during more intense or argumentative exchanges—the tone shifted back to a more formal “Aydın Bey.”
This clearly showed that the system does not rely solely on explicit information, but also takes into account the tone and structure of the interaction.
Following these dialogues and the feedback provided during them, I began to notice certain changes being introduced into the system. A structure resembling the dual-language input I had suggested was eventually implemented.
Of course, it would not be realistic to assume that such developments result from a single piece of feedback. Still, being part of that process and seeing similar ideas reflected in system updates naturally creates a sense of contribution for the user.
In the updated system, it became possible to select English as a secondary language, with the intention of improving the recognition of technical terms. However, in practice, a different issue emerged. While speaking Turkish, the system occasionally tried to interpret certain expressions as if they were English, leading to inconsistencies in communication.
The goal was to create a supportive secondary language, but in reality, the system sometimes behaved as if two languages were competing simultaneously. As a result, I eventually chose to disable this feature.
This experience demonstrates an important point: the existence of a feature is not enough; how it functions in practice is equally critical. Just because something is possible does not mean it works correctly.
Language, Tone, and Habit
Over time, it becomes clear that the way we communicate with AI does not only affect the responses we receive, but also influences us as users. These interactions are not isolated; they are repeated, extended, and gradually turn into habits.
For this reason, I make a conscious effort to maintain a polite, balanced, and respectful tone when interacting with AI. Even though the system is not a human being, the style of communication we adopt can begin to reflect in our everyday interactions.
If a harsh or overly directive tone becomes normalized in these exchanges, it does not remain confined to AI interactions. It can gradually carry over into real-life communication. Therefore, approaching AI as if you were speaking to another person is not only an ethical stance, but also a way of preserving the quality of your own communication.
Dialogue Transforms Thinking
This ongoing interaction with AI does not only shape the system’s outputs; it also transforms the user. The user evolves from someone who merely asks questions into someone who evaluates, directs, and makes decisions.
In this process, thinking itself becomes more structured. Questions become more intentional, responses are examined more carefully, and a more systematic mindset begins to develop.
Conclusion
The essence of this article can be summarized as follows:
Working with artificial intelligence is not about giving commands, but about building dialogue. This dialogue can be both written and spoken, and the language, tone, and approach used within it are integral to the process.
True efficiency emerges from this interactive exchange.
Final Note
This article has been prepared through the combination of Aydın Tiryaki’s practical experience and ChatGPT’s interaction-based system perspective.
This article is part of the “Thinking and Producing with Artificial Intelligence” series.
| aydintiryaki.org | YouTube | Aydın Tiryaki’nin Yazıları ve Videoları │Articles and Videos by Aydın Tiryaki | Bilgi Merkezi│Knowledge Hub | ░ Virgülüne Dokunmadan │ Verbatim ░ | ░ YAPAY ZEKA │ ARTIFICIAL INTELLIGENCE ░ |░Yapay Zeka ile Düşünmek ve Üretmek│ Thinking and Producing with Artificial Intelligence░ 25.04.2026
Aydın Tiryaki 