#25 Presenting with AI (1): Story — What Do You Want Them to Feel?

Someone is presenting in an online meeting. Half the cameras are off. One person is writing emails, another is reading something else. Afterwards, the presenter says what I hear more and more often: I would like some meeting rules. Cameras on, no email, full attention for whoever is speaking. I understand the wish. I also think the rules would not have saved that meeting. Many employees landed in a hybrid world of communication during Covid and never really learned how to communicate in it. There is more to do than fits into a day, and, I suspect, many of us have trained ourselves on thirty-second videos. But a good part of it sits with the sender: how you win a room's attention, and how you keep it.

If you run a company, this is your problem first. Decisions get made where someone presents: the board meeting, the sales call, the project review. In a Microsoft Research analysis of 34,524 meeting records, someone sent or answered email in roughly three out of ten meetings, and the odds were six times higher in meetings over eighty minutes than in those under twenty. The deck has become cheap since AI tools like ChatGPT and Copilot arrived. Attention has not. What the presentations in your leadership meeting are for is decided at the top, or by nobody.

(The first of three editions on presenting with AI. In my presentation trainings, when there is enough time, the revision runs in one order: content and story first, then the slides, then the person in front of the room. The editions follow that order, and one presentation grows alongside them.)

8 min read. If you take one thing from this: a presentation has two goals: what people should know afterwards, and what they should feel. The second one is usually missing. Ahead: data changes beliefs, stories change plans, you need both; the ending counts more than the middle, plan it first; read only your slide titles, are they the argument?; use an AI tool as a critic, not a fan, plus one task for whoever signs. Part 2 next: the slides, and what Copilot, Gemini, ChatGPT and Claude do to them.

Two goals, not one

Ask someone what their presentation is for and you usually get the first goal: what the audience should know afterwards. The second is what they should feel. Motivated? Worried enough to act? Curious to try something? That is the goal I want named, for the whole presentation and for each slide, and it is the one many decks never had.

The research supports the split more than I expected. A meta-analysis by Simon Zebregs and colleagues, then at the University of Amsterdam, compared statistical evidence with stories: numbers had the stronger effect on what people believed, stories on what they intended to do. Data moves beliefs. Stories move plans. A presentation that only informs can change what people think and leave Monday as it was.

The feeling has to be chosen per slide as well. Elizabeth Kensinger, a neuroscientist at Boston College who studies emotional memory, puts decades of work in one picture: after seeing a car accident on a street, people remember the accident well and the street badly. Emotion strengthens memory for the centre and weakens it for the edges. So one feeling per slide, attached to the one thing that slide is for.

And the ending needs the most care. Daniel Kahneman and colleagues showed in 1993 that people remember an experience by its peak and its end, not by its length: 22 of 32 participants chose to repeat the longer of two painful cold-water trials, because its last seconds were slightly less unpleasant. A presentation that fades out with a tired "any questions?" is remembered for the fading out. Stephen Kosslyn, a cognitive scientist then at Stanford, scored 140 public slide decks against psychological rules: none was free of flaws, and almost two thirds had no clear ending.

One line that holds it together

The red thread is a sentence. Chris Anderson, who ran TED from 2001 until this year, advises every speaker to put the through-line in fifteen words or fewer. If you cannot write it, the audience cannot find it. Once you have it, each slide serves that sentence or goes.

How the line is delivered depends on who is listening. Erin Meyer's Culture Map, an old companion of this series, has a scale for exactly this: some audiences want the conclusion first and the reasoning only on request (the US and the UK, in her placing), others want the principles and the method before a conclusion (France, Belgium, Spain; Germany nearer the middle). Her advice: "Presenting to Londoners or New Yorkers? Get to the point and stick to it. Presenting to Belgians, Spaniards, or Germans? Spend more time setting the parameters and explaining the background before jumping to your conclusion." One qualification, as always: roughly eighty percent of the variation in cultural values lies within countries, not between them. Nationality is a weak clue; the people in the room are a stronger one.

If you sign budgets and the presentations in your leadership meetings end without anyone knowing what to decide, one thing is enough. Forward this edition to whoever runs people development for you, with one line: "I want us to work on this over months, not in an event. Find me something like it." What to look for: a programme that starts with the leadership team, runs over months rather than days, and treats human skills and AI as one subject. If you run people development and just received that line, or still have budget for this year: that is the work I do, since 2010 with organisations from start-ups to adidas and Fraunhofer (references). Twenty minutes, and you'll leave with a clear next step whether or not you work with me: book a call.

What goes on the slide, and what gets said

The first rule: the title of a slide is a sentence, not a topic. Not "Q3 results" but "Q3 grew in two markets and lost the third." Michael Alley, who teaches engineering communication at Penn State, and Joanna Garner of Old Dominion University tested this with 110 students: the same eight-minute talk with sentence titles and one visual piece of evidence per slide, against the usual topic-plus-bullets version. Comprehension was clearly higher straight away, and recall still better ten days later. The test you can run tonight: read only your titles, top to bottom. If they are the argument, the deck works. If they are a table of contents, not yet.

The second rule: the script does not belong on the slide. Richard Mayer, an educational psychologist at UC Santa Barbara who studies how people learn from words and pictures, found in sixteen out of sixteen experimental comparisons that showing the spoken words as text on screen makes people understand less, not more. Kosslyn's survey of 205 people found the same from the audience side: the top annoyances included speakers reading their slides word for word, and the main point lost in detail. The slide carries the claim and the evidence. You carry the story and the feeling.

Where AI helps in step one

Here the tools help, on one condition: as a critic, not as a fan, because the models tend to please. Researchers at Anthropic showed in 2023 that five leading AI assistants lean towards agreeing with the user rather than correcting them, and in April 2025 OpenAI rolled back an update to ChatGPT because it had become, in their own words, "overly flattering or agreeable". Ask "is my presentation good?" and you will probably hear that it is. Ask differently and they are useful: a Stanford team led by James Zou gave GPT-4 more than 4,800 research papers to review, and its comments overlapped with human reviewers about as much as two human reviewers overlap with each other.

So the prompts that work in step one are the hostile ones. Paste the titles only: "What is the argument of this deck, in one sentence?" If the answer is not your through-line, the deck does not say what you think. Then: "Which slide would a finance director stop listening at?" Then, per slide: "What should the audience feel after this one?" If the answer is "informed", the slide has one goal. One limit: a model playing your CFO plays the average CFO, not yours.

I am applying these prompts to the presentation that grows across these three editions: the case for investing in human skills and AI together, now, to stay competitive, made for a management board. Through-line, fifteen words: "Anyone can buy our AI tools. The difference is how our people use them together." The feeling at the end: determined, and a little impatient. Below is the working slide for the titles test: five titles, one feeling each.

And the last slide, written first: a decision with a date. The feeling, resolve, is not on the slide. You carry it into the room.

The rest of the deck, and how the tools built it, next time.

Why this step stays yours

The deck was never the point; the room is. Vanessa Bohns, a social psychologist at Cornell who studies how people ask for things, found with Mahdi Roghanizad that requests made face to face were granted many times more often than identical requests by email, while the people asking expected both to work. In two follow-up experiments with over two thousand people asked, the order held: in person, then video or phone, then email. People believed a video call would do as well as the room. It did not.

Two newer findings add to this. A Stanford study of 1,601 people found that labelling an argument as AI-written did not make it less persuasive. The words are not the problem. But Oliver Schilke and Martin Reimann at the University of Arizona showed in thirteen experiments that people who disclose using AI are trusted less, in their examples by sixteen to twenty percent. The text can come from anywhere. The trust stays with the person in front of it. In the AI-text edition I wrote about Denmark, where from January pupils in one branch of upper secondary school must defend their large home-written assignment out loud: "You can hand in a text you didn't think through. You cannot defend one." A presentation is that defence.

Your turn

If you sign budgets. Before the next board deck, ask the presenter two questions: what should we decide, and what should we feel? Then send this edition to whoever runs people development, with the line from the middle of this page.

If you lead a team. Read only the titles of the next deck that reaches you. If they are a table of contents, send it back with one request: make each title a sentence.

If you develop people. Before booking another slide-design training, run a rehearsal in which every slide has to name its fact and its feeling out loud. Some slides will not survive it. If you need the mandate, forward this upward; the opening is written for the person who signs.

Your 1% this week, whoever you are. Write the through-line for your next meeting in fifteen words or fewer, before opening the slides.

Next time: the slides, and what Copilot, Gemini, ChatGPT and Claude do to them. Until then: what did the last presentation in your leadership meeting want you to feel? Tell me in the comments. I read every reply.

If you want to bring these themes into your organisation, here's that link again: book a 20-minute call. You can also explore the Leadership Essentials Workshop Series for teams.