What Happened
Researchers published a study examining how AI systems shift their communicative character based on minimal input from users. The paper, released in Frontiers in Artificial Intelligence, found that small changes in prompting can measurably tilt an AI response toward logic, emotion, or practical guidance. Scientists now have data confirming what experienced communicators already knew: word choice shapes everything, even in machines.
The Communication Angle
This research is a mirror held up to human communication, and most people won't like what they see.
Here is what the study actually proves: tone is not decoration. It is the message. When researchers adjusted a handful of prompted words and watched the AI's entire emotional and intellectual posture shift, they demonstrated something I have been teaching for years. The words you choose do not describe your intent. They ARE your intent. The listener, human or machine, receives the architecture of your thinking through the specific words you select. Change the words, you change the thought itself.
Most professionals treat tone as an afterthought. They build their argument, load it with facts, then sprinkle in a word like "unfortunately" or "excited" and consider the emotional dimension handled. That is backwards. The AI research shows that communicative character, the balance of reasoning, warmth, and direction, emerges from the ground up, from the smallest word choices. It is not a layer you add at the end. It is the foundation.
Here is the practical problem this creates. If a few prompted words can redirect an AI's entire personality, imagine what your habitual word patterns are doing to your professional relationships. The colleague who always frames feedback as a question. The manager who opens every difficult conversation with "I just want to say." The executive who ends every directive with "does that make sense?" Each of those habits broadcasts a communicative character. Tentative. Deflecting. Uncertain. And the audience reads it whether you intend it or not.
The researchers measured what they called the "titration" of AI voice. I want to borrow that word because it is precise. Titration means adding something drop by drop until you hit the exact balance you need. That is what skilled communicators do. They do not dump emotion into a message or strip it out entirely. They calibrate. One specific word added here, one removed there, until the message carries exactly the weight and warmth the moment requires. The AI study shows this is not intuition. It is a measurable, learnable craft.
This is exactly the kind of scenario I break down in Say It Right Every Time. The chapter on tonal calibration gives you a framework for auditing your own word patterns the same way these researchers audited the AI. You will find specific word substitution drills that show you how swapping one word in a sentence shifts the perceived confidence, warmth, or authority of the entire message. The AI just made it visible. The book makes it actionable.
Never Be Lost
for Words Again
Get word‑for‑word scripts for the conversations that shape your life, from job interviews and negotiations to difficult talks with family and partners, so you always know exactly what to say and how to say it.
Never Be Lost
for Words Again
Get word‑for‑word scripts for the conversations that shape your life, from job interviews and negotiations to difficult talks with family and partners, so you always know exactly what to say and how to say it.
Key Takeaway
Before your next important email or presentation, identify the single emotional register you need the audience to receive: confidence, urgency, reassurance, or authority. Then read your draft and circle every word that contradicts that register. Replace those words first, before you change anything else. Do not rewrite the whole thing. Just eliminate the contradictions. That one edit will do more for your communication than restructuring the entire message.
