AI and Communication: The Insidious Impact on Our Human Interactions

You Talk to Your AI Like It's a Slave

Daily interaction with artificial intelligences (AIs) subtly transforms our behaviors, influencing how we communicate with our colleagues. When addressing an AI, we often adopt a directive, almost imperative tone, which eventually spills over into our human relationships. This phenomenon is fueled by the fact that AIs obey without resistance, reinforcing a reflex that extends beyond the screen.

The Prompt Has Replaced Conversation

A few years ago, we approached AIs with politeness, using courteous phrases like "Could you help me with..." or "Would it be possible to...". However, over time, we learned that direct commands were more effective. A study from Penn State in 2025 showed that this directive tone improves the accuracy of AI responses by 4 percentage points on complex tasks. This optimization method, while beneficial for AI efficiency, has unintended repercussions on our human interactions.

Operant Conditioning and Acquired Reflexes

The concept of operant conditioning, illustrated by BF Skinner in the 1950s, explains that when our behavior produces a reward, it automatically strengthens. Thus, by using a directive tone with AIs and receiving more accurate responses, we reinforce this behavior. This phenomenon is not without consequence, as it also alters how we communicate with humans. Our brains, through neural plasticity, adopt the most frequently used register as the default, meaning that the terse commands directed at AIs find their way into our daily exchanges with colleagues and loved ones.

Your Brain Doesn't Make the Distinction

Communication habits transfer from one context to another. The language patterns we develop with machines influence how we speak to humans. This phenomenon is due to neural plasticity, which causes our brains to adopt the most frequently used register as the default. Thus, the terse commands directed at AIs appear in our daily exchanges with colleagues and loved ones.

A musician who practices a piece thousands of times ends up playing it in their sleep. A salesperson who uses the same pitch for ten years also uses it with friends. A manager who spends four hours a day issuing terse commands to a machine... you see where this is going.

The Continuous Training of AI Models

Large AI models, such as ChatGPT, are not static. They are continuously retrained on our conversations. OpenAI's terms of use specify that each interaction feeds into the training of future models. Research from Anthropic in 2026 revealed that these models incorporate emotional markers that influence their responses. Thus, the AI of tomorrow will be shaped by how we speak to it today, and by extension, by how we interact with each other. This means that AI will become more directive, more commanding, and less nuanced, reflecting what we have sent it, amplified and normalized.

Technostress 2026: A Silent Reconfiguration

The phenomenon of technostress has evolved since the 2010s, when it was primarily associated with email overload and endless Zoom meetings. By 2026, technostress describes the silent reconfiguration of our relational patterns through intensive daily interaction with entities that lack human relationships. A machine does not get offended when you are terse; it does not expect reciprocity; it simply executes. This constant interaction with entities devoid of human emotions unlearns our tolerance for others' imperfections, patience, and the ability to formulate requests that allow space for others to exist.

A machine does not get offended when you are terse. It does not need to be asked how it is doing. It does not expect reciprocity. It executes. And that is exactly what makes it dangerous for you.

What This Says About Us

There is a word for this phenomenon: technostress. Not the version from the 2010s — email overload and endless Zoom meetings. Technostress 2026: the silent reconfiguration of our relational patterns through intensive daily interaction with entities that have none.

Because by interacting for hours a day with an entity that operates this way, you unlearn something essential: tolerance for others' imperfections. Patience with someone who needs context. The ability to formulate a request that allows space for others to exist.

What's at Stake: QVCT as the New Battleground

Let’s return to the initial experience. Your last ten messages to your AI. Your last ten messages to your colleagues. If the tone is identical, you have an immediate operational problem: your collaborators are not LLMs. They do not respond better to terse commands. They respond worse. They become demotivated, withdraw, lose confidence. And unlike your AI, they remember. Everything.

But this is not just a management issue. It is a workplace health issue. Psychosocial risks — PSR in the vocabulary of HR and prevention specialists — arise precisely from this: degraded professional relationships, interactions stripped of their human dimension, a growing sense of being treated as a tool rather than as a person. The INRS has documented this for years. What no one anticipated was that the source of this degradation could be... a browser tab open in parallel.

Quality of Life and Working Conditions — QVCT, to use the framework of ANACT — rests on a simple foundation: work relationships where humans remain at the center. Not as a slogan. As a daily operational reality.

This foundation is being eroded. Not by malice. By the invisible conditioning of a technology we use without measuring what it does to us.

You cannot prompt trust. And you cannot deploy a serious QVCT approach in an organization where managers have learned, tool by tool, interaction by interaction, that the other does not need to be treated with consideration to execute.

Your AI will not hold it against you if you talk to it like a slave. It has no self-esteem. No memory. No tomorrow.

Your colleagues have all three. And in three years, when AI models have been retrained on hundreds of billions of interactions that are predominantly directive, impatient, and dehumanized — when the standard AI has become the amplified mirror of our collective way of interacting — the real question will no longer be technical. It will be anthropological.

Have we learned to talk to a machine? Or have we unlearned how to talk to humans?