Output is not yet an outcome: Why AI texts do not work automatically
A linguistically sophisticated text is not yet an effective text. Output is what we produce. Outcome is what we achieve with it. For communication, this means: It is crucial whether our words create understanding, strengthen trust, or motivate people to take action.
A new study on AI-powered petitions makes this distinction particularly clear. For me, it raises a question that goes far beyond writing: If convincing-sounding words are available at any time—what else do we have to offer ourselves?
What the study on AI writing assistants shows
Isabel Corpus, Eric Gilbert, Allison Koenecke and Mor Naaman examined around 1.5 million petitions on Change.org to see what changed with the introduction of an integrated AI writing assistant.
Your result: Access to the AI tool changed the linguistic characteristics of the petitions and made the texts more similar to one another. However, the success measures examined did not improve at the platform level. The study was published on September 30, 2026. Nature Human Behaviour. About the original study
This finding does not mean that AI texts are fundamentally ineffective. It shows for the examined use case on Change.org: Changes in the linguistic form did not automatically lead to better results. Whether the lack of a personal voice was responsible for this cannot be inferred from this.
Why this finding reminds me of my work with Charisma
For decades, I have been studying how people exert influence through their personalities. From this work, I know of a similar challenge: A presentation can be rhetorically sophisticated and still fail to achieve its goal.
The structure is correct. The phrasing is appropriate. The tools are properly used. Yet the question remains: Is the person feeling the impact of their request?
In my experience, credibility is also created when words, attitude, and behavior are aligned. Therefore, I am particularly interested in the study of the distinction between an optimized outcome and its actual effect.
My conclusion for practice: Linguistic optimization should make one's own perspective clearer.
How one's own voice remains recognizable in an AI text
AI can help with organizing, summarizing, and formulating. I have to decide for myself what statement I want to make, what experience it brings, and for what I take responsibility.
For that, I have three questions:
- What specific observation does my statement make?
A situation from my work or my life shows what my assessment is based on. - What attitude do I represent?
What is important to me about this topic — and why? - What is the text supposed to achieve for its readers?
Are they supposed to understand something, question a decision, or take a concrete step?
An example: „Authenticity is important“ remains abstract. Anyone who describes a concrete situation in which words and behavior matched or clashed makes the statement understandable.
That requires our own contribution: observation, judgment, and the willingness to take a position.
Charismatic influence in the age of AI
My Charisma concept focuses on the personality. Therefore, my guiding question when using AI is:
How will you recognize me in this text?
Based on my experience? My values? An observation I made myself? Based on a statement I stand by?
For me, it is about making the whole of oneself visible through all the linguistic care. I also devote myself to this topic in the expanded edition of my book. „Charisma: Shape your life. Change the world.“
The cover image reminds me of this attitude. It was created during the premiere evening of To PRACTICE by Laurens Pérol. The film tells the story of a musician who, in accordance with her values, sets off on foot to Oslo for her first rehearsal.
This leads me to a personal question:
What own contribution am I prepared to make so that the result corresponds to my values?
This question accompanies me when writing, in conversation, and on stage. In the age of AI It becomes even more important for me as a result.
