Recognising AI authorship in public speaking

By now, you have heard it. You have heard all of it. Twice. Once at last year’s conference, and again — word for word — at this year’s.

It always begins the same way. “In today’s fast-paced, ever-evolving world of unprecedented challenges and limitless opportunities…”

Cue the dramatic pause. Another pause. Then another.

By this point, even the audio technician is mouthing along with the speaker.

Then come the adjectives, those poor, overworked words dragged out like exhausted marathon runners. Words like transformative, empowering, disruptive, paradigm-shifting, unprecedented. (again — because it would not be a proper AI speech without a bit of recursion).

We are living in the golden age of ChatGPT Oratory. The speeches are everywhere. Boardrooms, conferences, convocations, TedX talks, even weddings. Anywhere there is a microphone, there is someone earnestly delivering a perfectly structured, syntactically impeccable, spiritually vacant speech.

It is no longer plagiarism we need to fear. It is monotony. No two speeches may be identical (thank you, token randomness setting), but they are certainly siblings. Half-brothers raised by the same overworked LLM, taught to speak in bullet points, crescendos and faux gravitas.

The tell-tale signs of AI authorship:

“I stand before you today…” (Thank you for confirming your verticality. We were concerned.)

Tired metaphors about journeys, mountains, bridges and doors we must unlock to discover our limitless potential. Occasionally a lighthouse or a phoenix if the user was feeling adventurous.

The sudden, deeply unearned appearance of words like zeitgeist, serendipity or synergy.

An awkward AI-generated joke wedged in because someone typed “make it light-hearted.” (It was not. It never is.)

And of course, the pauses. Those…long…pauses… Because somewhere in OpenAI’s dataset is a TED Talker who made dramatic pausing sound important.

By now, even the water bottles on the podium know when to brace for impact.

One of my corporate friends talked about a recent townhall. During the event, a senior VP earnestly delivered a speech about “navigating the turbulent seas of digital transformation.” His team knew instantly it was AI-generated — because the VP has a pathological fear of water and once got seasick on a pedal boat.

One brave intern pointed this out afterward. The VP blinked slowly and said, “I just thought it sounded visionary.” The intern was fired. The speech template lives on.

At a wedding I attended a few months ago, the groom’s best friend delivered a speech allegedly “from the heart.” However, somewhere between the phrases “unlocking the doors of tomorrow’s shared happiness” and “resilience in the face of marital headwinds,” it became obvious that ChatGPT5 had inspired this heart.

The truly comic moments, though, are when ChatGPT hallucinates. Midway through an otherwise bland speech about leadership, the speaker solemnly announces:

“As Aristotle said in his TikTok channel just last week…” Or confidently

attributes a quote to “Winston Churchill’s podcast.”

The audience, starved for originality, laughs. Not at the joke (there was not one), but at the absurdity of watching a human being become a ventriloquist dummy for a large language model’s dream. Meanwhile, the speaker, blinking in confusion, wonders why they are laughing at what was supposed to be the inspirational climax.

Perhaps the greatest irony? People using AI to sound more human have ended up sounding more machine-like. Uniform. Predictable. Comfortably bland.

The next frontier is not AI detection software — it is AI group therapy for audiences who cannot endure another keynote about resilience, innovation and unlocking the future.

If you are lucky, maybe next time, two ChatGPT-written speeches will collide head-on at the same event, delivering near-identical closing lines:

“Together, we can shape a better tomorrow… starting today.”

Cue standing ovation. From the AI bots. The humans have already left the room.

Author’s Note: This article was written without the assistance of ChatGPT. However, several AI-written speeches were harmed in the making of it.

This post was published on October 21, 2025 7:38 pm