AI-written talks: when the audience starts to doubt your talent

At a company's annual convention, an executive who opens a talk by mentioning the "challenges of a constantly changing market" and goes on for twenty minutes with polished sentences and three pillars neatly spread across the slides runs a silent risk. During the coffee break, the audience may comment in the team's WhatsApp group that the opening seems to have come out of ChatGPT. The person on stage almost never hears about that conversation, and from then on every sentence for the rest of the day is heard with a question in the background: how much of this belongs to the person speaking?

We begin this series with that moment. It happens in silence, far from the stage, and it hits precisely the people who have the most to lose. That may be your case: someone who got there through talent, preparation and years of results.

AI-written talks and the credibility of the person on stage

When you go on stage at a convention, a conference or a launch, you lend your own name to the speech. The audience evaluates the content and, along with it, evaluates you: your preparation, your experience, your care for the people sitting there. If you lead a department or have been giving talks for a long time, you built that reputation over years of meetings, negotiations and presentations that went well. That entire asset is exposed when someone in the audience starts to doubt where the text came from.

Doubt usually comes from small details. A catchphrase that would fit any talk. A story with no name, no place and no date. A conclusion that repeats the opening in other words. For listeners, the question alone already carries weight. Cornell researchers showed, in a study published in 2023 in the journal Scientific Reports, that the suspicion that someone used AI suggested replies in a conversation, even when wrong, is enough for that person to be rated worse and seem less cooperative. According to the Cornell Chronicle, people come to evaluate the other person "more negatively simply because they suspect you are using AI".

In our view, this means that for anyone who gives talks, the risk exists before any proof. Your real talent, what you actually thought and lived, falls under the same doubt as the generic paragraph.

The image cost of using AI, according to research

In 2025, researchers Reif, Larrick and Soll, from Duke University, published in PNAS a study with four preregistered experiments and nearly 4,500 participants. In one of them, with 1,215 people, those who used AI at work were rated as "lazier, less competent and less diligent". The effect appeared regardless of age, gender and profession. The samples were collected online and measure first impressions. The fourth experiment brings a nuance you need to know: the penalty disappeared when AI was clearly useful for the task.

At the University of Arizona, Schilke and Reimann ran 13 experiments with more than 5,000 participants on what happens when someone says they used AI. Trust fell 16% for a professor who grades with AI, 18% for companies that disclose AI in ads and 20% for graphic designers. Softening the disclosure with expressions such as "just proofreading" did not prevent the drop. It was even greater when a third party exposed the use, a situation the university's website describes as "the worst possible outcome when it comes to trust". Remember the comment in the WhatsApp group at the beginning? It looks a lot like that third party.

Coman, from the University of Florida, and Cardon, from USC, surveyed 1,100 professionals about work messages written by managers. The perceived sincerity of a manager who uses a lot of AI was between 40% and 52%, against 83% for a manager who uses little. Professionalism was between 69% and 73%, against 95%. The penalty was greater in emotional or congratulatory messages, persuasive messages and performance reviews. Think about your next convention talk: it will probably be persuasive and, quite often, emotional.

Toastmasters and Nancy Duarte on AI-written talks

In April 2026, the official magazine of Toastmasters International published an article by David Tobenkin on using AI to write speeches. The central line is direct: "AI should not write your speech." The same text recommends that any AI generated content be reworked until it is closer to how the person thinks and speaks. In 2023, another article in the same magazine, by Victor David, already predicted that speeches written by AI "will, unfortunately, sound artificial" and that the organization's contest judges would probably take off points for lack of originality.

Nancy Duarte, from Duarte Inc., wrote in April 2025 that AI can make a presentation "efficient and empty". Further on, she states: "Speed replaces empathy, and your message won't resonate." In the same text, Duarte treats AI as a collaborator and accepts using the tool. For her, every great presentation starts with empathy for the people who will listen.

The mockery Léo talks about happens in the messaging group, at the lunch table, in the car on the way home. Always far from the person who was on stage.

Your talent called into question

The person with the most to lose from a generic talk is precisely the good professional. If you made it to the convention stage, you made it on merit: you made tough decisions, built teams, achieved results. That track record is the most valuable material you have to offer an audience, and it is exactly what a generic text leaves out. When the speech sounds like any other, the audience hears the tool's voice over your biography.

Answer frankly, just for yourself:

• When was the last time someone quoted a line of yours after a talk?
• Did anyone in the audience come up to talk to you about a story you told on stage?
• If another director read your text, would it work for their company with almost no changes?
• Can you explain the central idea of your last presentation in two sentences, without looking at the slides?

If you are a master of ceremonies or an event protocol professional, you face the same risk in a format of your own. In a ceremony, every name, every thank you and every tribute carries the reputation of the entire event, and a generic text there appears in front of all the guests at once.

What the Talks and ceremonies series will show

Over twelve articles, this series brings together studies, official guidelines and cases documented by the press about talks, keynotes, corporate presentations and ceremony scripts written with generic AI. The second article brings what great speakers and experts say, such as Chris Anderson, Nancy Duarte, Carmine Gallo, Julian Treasure and Brené Brown. The third details TEDx guidelines on your own voice, rehearsal and reading on stage. The fourth presents the competence penalty measured by research, and the fifth discusses the transparency dilemma: whether or not to tell the audience that AI took part in the text.

In the sixth article, you will find cases of events and speeches that turned into crises. The seventh deals with the master of ceremonies and the ceremony script, where a mistake is visible to every guest. The eighth examines Toastmasters guidance, and the ninth shows the value of your own story, the material no tool has. The tenth gathers what the audience expects from a keynote. The eleventh explains how to use AI in preparation without losing credibility, and the twelfth shows how working with a scriptwriter for talks and ceremonies works.