I've Assigned Hundreds of Student Presentations.

Now that AI can make the slides, I'm not exactly sure what they're for

I confess that I am a slide deck nerd.

I own almost a dozen books on presentations - Nancy Duarte’s Slide:ology, Garr Reynolds’ Presentation Zen, Carmine Gallo’s Presentation Secrets of Steve Jobs, to name just a few.

I’ve watched Steve Jobs introduce the iPhone more times than I care to admit, knowing how much he rehearsed in front of slides mostly containing a single phrase, chart, or image while delivering one of the most consequential product announcements in history.

Look at the table of contents in any of the books referenced above. Duarte devotes chapters to “Creating Ideas, Not Slides,” “Thinking Like a Designer,” and “Arranging Elements.” Reynolds structures his book around Preparation, Design, and Delivery, with chapters like “Planning Analog,” “Crafting the Story,” and “The Art of Being Completely Present.” Gallo stresses the importance of questions, roadmaps, numbers, and “amazingly zippy” words. These aren’t books focused on formatting or fonts - they’re books about thinking, storytelling, and communication.

I know what good presentation design looks like using key principles such as simplicity, storytelling, and the rule of thirds.

I also violate them constantly. And I suspect I’m not alone.

Not out of ignorance, but out of time. Creating a truly well-crafted presentation - actually storyboarding your ideas before opening your laptop, curating the perfect image that captures your message, maintaining the discipline of consistent formatting - all of that can take hours that frequently turn into days.

Teachers everywhere face the same hurdle. Despite knowing how difficult it is (or perhaps because of it?), we frequently require students to create and deliver presentations, often as part of a “group” project. Millions of students around the country will dutifully create Google Slideshows or PowerPoint presentations again this year.

It’s worth asking: when we assign students to make presentations, what exactly are we expecting them to do? What are we evaluating? And what skills are we actually teaching? What are first principles when it comes to in-class presentations?

The uncomfortable question in 2026: now that AI can create professional slides instantly, does that change the equation?

The Capability Jump

Every major AI platform - ChatGPT, Gemini, Claude - is currently capable of creating individual slides and even complete slide decks in seconds. Gemini’s Nano Banana debuted last fall and seemed to solve the problem marrying accurate text and image generation, spawning a million infographics overnight.

Other tools, like Gamma, are entirely integrated with AI-powered presentation features. Just as we’ve been concerned about cognitive offloading with student writing, educators now need to be far more intentional about the purpose of assignments that require student presentations.

Students can upload their notes and immediately receive back polished slides that look better than most of us could produce in a week. No more fiddling with fonts, colors or layouts. No more searching for useful graphics or images that visually convey a complex concept.

This isn’t hypothetical. It’s happening now, in classrooms everywhere. And it forces a question that many teachers have not even considered: if AI can handle the visual design, the formatting, the organization of slides - what exactly is left for students to do?

I Know What I Hate

Here’s another confession from a veteran teacher: Most slide decks are terrible.

I’ve assigned hundreds of presentations over the decades, and I still don’t know precisely what I’m after.

What I know is what I hate. Text-heavy slides. Students reading bullet points verbatim. Images that have little to do with the content. Presenters who clearly don’t understand their own material. Worst of all, the interminable group presentation where three or four students take turns reciting dense paragraphs on the screen with their backs turned to the class.

But knowing what’s bad isn’t the same as knowing what’s good. And knowing what’s good isn’t the same as knowing the skills I’m trying to teach.

When I assign a presentation, am I assessing:

• Visual design quality? The ability to create aesthetically pleasing slides with appropriate balance of content and imagery?

• Information organization? The skill of structuring content into a logical flow?

• Technical proficiency? Knowing how to use PowerPoint, Google Slides or Keynote?

• Research and synthesis? The deeper work of gathering information and distilling it into key points?

• Public speaking and communication skills? The ability to explain ideas clearly to an audience with confidence, presence, and the capacity to think on your feet?

Even if I want to answer yes to all of them, my internal rubric about a strong presentation is not much more accurate than Justice Potter Stewart’s famous maxim about obscenity - I know it when I see it.

The list above covers very different learning objectives. And I’d wager that, like many of my colleagues, I’m primarily interested in content mastery. The visual quality of the slides is far less important than whether the student understands the causes of the Civil War or the failures of Reconstruction. Even the most beautiful slides won’t tell you whether that’s the case. A well-designed slide show is a bonus, but in the absence of true comprehension, it’s not evidence of learning.

If that’s true, why do we allot so much time for students to create slide decks?

Why Slides Feel Different from Writing

Here’s something I’m also wondering: are teachers more willing to allow students to use AI for presentation assistance than with writing? If so, what might be the reason?

Maybe it’s because we’ve always known slides were somewhat fungible. Thousands of templates exist and borrowing layouts is normal. I’ve often asked students where they found a visually arresting background after their turn at the SmartBoard. Nobody accuses a student of cheating because they used a pre-designed Google Slide theme. The slide itself was never the point. It was always a delivery mechanism for something else.

Pure writing assignments feel different. An essay is a direct connection to you - your thoughts and your voice. But a slide deck? Slides, especially the way most are constructed in K-12, feel more like supporting material. And slide presentations are almost always accompanied by an oral component which can serve as a backstop to gauge understanding. This asymmetry might explain why the AI-and-presentations conversation hasn’t generated the same panic as AI-and-essays.

But there’s a catch, isn’t there? After all, slides contain writing! And, just like with regular writing, AI can generate presentation content from training data without the student ever doing the reading. The student who uploads a textbook chapter to an AI and gets back a summary deck hasn’t learned anything. The only failsafe is the presenting itself - which is why that’s where our assessment focus needs to be.

Of course, like everything else with AI, no assignment is bullet proof for students determined to outsource their thinking. They can still generate AI scripts and I’ve written elsewhere about how this plays out in debate, where live performance backfires when students read text they didn’t create or review.

The answer points in the same direction: a formal Q&A period. Asking students to explain their reasoning, integrate facts, and connect their ideas to the content needs to be a component of any assignment which involves a slide show. A student reading a memorized script - whether group-written or AI-generated - will struggle when pushed beyond the prepared text if they don’t understand what’s actually contained in their presentation.

The Uncomfortable Question

Here’s a genuinely uncomfortable question with respect to the slides themselves:

If a student uploads their notes to an AI platform - notes they took, from research they conducted, reflecting ideas they developed - and it produces professional looking slides, and then the student stands up and explains the material cogently, answers questions thoughtfully, demonstrates clear understanding through their delivery...

What exactly has been offloaded?

The visual design? Absolutely. But if graphic design isn’t what I’m actually trying to teach - and for most K-12 teachers, it isn’t - then what’s the problem? How many assignment instructions include principles like the rule of thirds, the importance of visual storytelling, or the power of using tables, graphs and quotations?

Are we worried about AI usurping slide organization? Figuring out the correct order in which to present information is definitely a skill. But is it a skill most teachers explicitly teach and evaluate? I might point out where a map would have been a useful visual anchor, but I rarely penalize a student for that kind of lapse unless I’ve made it an explicit requirement in the assignment.

In other words, if AI structures a student’s notes that were already organized - if the student already knew what points they wanted to make and AI just decided a map would be helpful on slide three - is that a detrimental cognitive offload? Maybe. Or is it a useful piece of visual feedback to the student that a map is an ideal image for that slide?

For me, what I care most about is the presentation itself which is the part AI can’t do. The explaining, speaking, and question answering. A strong oral presentation with no slides at all is far superior to one where a student has beautiful images and layouts but no idea what the presentation is about.

But What About the Struggle?

There are plenty of reasonable objections to allowing students to generate AI slide shows: having to make slides yourself forces a certain kind of thinking.

When you have to decide what goes on each slide, you’re forced to prioritize. When you have to find an image, you have to think about what visual would actually communicate your point. When you struggle with figuring out what you need to share, you’re confronting the gap between what you want to say and what you can show. That friction may be pedagogically valuable.

But my response to teachers who immediately cry foul when it comes to using AI generated presentations is that these arguments only really land if we’re actually teaching those things. If we’re scaffolding the design thinking, walking students through the principles Duarte, Reynolds, and Gallo write about, and demonstrating how to storyboard before they touch a keyboard.

In my experience, most of us aren’t doing that. We’re just assigning students to create slideshows and hoping for the best. If you’re not actually teaching the skill, you can’t claim students are learning it through struggle. They’re just struggling. A more truthful observation is that slideshow creation is often the province of “group work” where students are frequently left to their own devices to create their slideshow during a work period.

And for those teachers who do zero in on the quality of the presentation - how many teach storyboarding or the intricacies of visual design? My guess: for most of us in the business of prioritizing content, that’s not where we want to spend our time.

There's a deeper question here too: if AI has already internalized the principles taught by the experts - the rule of thirds, visual hierarchy, one idea per slide - do students need to learn them at all? Or is that knowledge now simply embedded in the tools? I'm genuinely uncertain. We clearly lose something when we outsource any skill. But given how terrible most slideshows actually are, it's worth asking whether better presentations via AI might be a reasonable trade - so long as the students still do the heavy intellectual lifting.

What This Looks Like in Practice

I recently experimented with this in my advanced Government and Politics class. I showed students the infographic feature within Google’s NotebookLM. I supplied source material and instructions on prompt design. Students were able to create detailed infographics on important legal principles related to the Bill of Rights derived from the Fourth, Fifth, Sixth, and Eighth Amendments.

You’ve seen this kind of output all over the web. Many AI-generated infographics are overwhelming - too busy, text-heavy, overstimulating. But with refined and iterative prompts, students were able to create visually striking, useful single-slide summaries.

What struck me was where the intellectual work actually happened. NotebookLM's infographic feature doesn't let you edit the output - what it generates is what you get. That meant students had to think carefully before prompting: What content should I feed it? What do I want emphasized? The students who produced the best infographics weren't the ones who knew the most about design. When the output missed the mark, they had to diagnose why and revise their approach.

They were the ones who understood the content deeply enough to know how to explain their slide to the class. These were students who had carefully read the material, took notes on the details, and were able to articulate how the principle fit within the purpose of the Amendment.

This is crucial. AI slides can quickly slip into doing all the cognitive labor if there is no requirement of up front engagement with the actual source material which includes careful reading, note-taking, and time for questions.

The slide itself then merely serves as a visual device to frame their short presentation for the rest of the class.

That’s where the understanding became visible. A student could have a strong infographic, but if they couldn’t explain why we have a “right to counsel” or why it mattered, it was immediately obvious.

The Rubric Challenge

Here’s a thought experiment for teachers. Pull out your presentation rubric. (If you don’t have one - and many of us don’t - you’re likely not alone.)

Look at what you’re assessing. What percentage of points goes to things AI can now do, such as visual design, slide formatting, organization, or other superficial elements?

What percentage goes to things only the student can do? Oral explanation. Responding to questions. Demonstrating genuine understanding through live delivery.

My guess is that most rubrics - if they exist at all - primarily assess the content of the presentations. Are the bullet points accurate? Do the slides cover the assigned material? That makes sense: as subject matter experts, it’s what we care about most, and easiest for us to assess.

But here’s the problem: the visual content can now be generated. The presentation itself - the live human performance - cannot.

Have we been assessing the wrong things when it comes to student presentations? Maybe not wrong exactly. But, in my experience, I’m definitely guilty of conflating multiple skills that I should have been evaluating separately.

What Would It Look Like to Do This Right?

If AI can handle the slide production, what if we made the preparation work the assignment?

Reynolds talks about “planning analog,” working away from the computer first, sketching ideas on paper, storyboarding the flow before touching any software. Duarte emphasizes “creating ideas, not slides” and the intellectual work that precedes any design work. Both books spend chapters on “crafting the story” before a single word appears on screen.

What if that became the assignment?

Imagine asking students to submit a written storyboard with the arc of their presentation mapped out. They'd include notes explaining what each slide needs to accomplish, an articulation of their audience and purpose, and a rationale for which visual elements would best communicate their key points.

Then - and here’s the provocation - what if we let them use AI to generate the first draft of slides from that preparation work? Make them revise the AI output based on what they actually want to say. And then assess them on the delivery.

Is this surrendering to AI? Or might it be using AI to clarify what we actually care about: that students can think through a complex topic, organize their ideas for an audience, and explain those ideas clearly to other people?

The slide production could become a step in the process rather than the end point. The understanding - demonstrated through in class explanation - is what remains paramount. I’d much rather have a class with students drafting with colored pencils and paper than spending 15 minutes trying to adjust digital images to fit inside of a slide template.

Where Does This Leave Us?

I don’t have a neat conclusion, but in the spirit of first principles, I have a set of questions I’m now forced to take seriously:

When I assign students to create a presentation, what am I actually trying to teach?

If AI can handle the visual design - and handles it better than many students could do on their own - is that a problem or a gift?

Should I be grading the slide presentation at all, or only the delivery and Q&A?

What would my rubric look like if I designed it primarily around the skills AI can’t replicate?

Maybe the presentation assignment was always really just about presenting. The slides were visible artifacts but the actual thinking could only be evaluated during the presentation itself. If that’s true, then AI-assisted slide creation might actually clarify what we were after all along.

Or maybe I need to admit I never really knew what I was teaching when it came to slide presentations. I knew what I hated - death by bullet point, mumbled delivery, and visual incoherence. I imagine this is what was going through Jeff Bezos’s mind when he famously banned Powerpoint from meetings and introduced the 6-page memo.

But Steve Jobs’ famous MacWorld Keynotes and thousands of Ted Talks support the idea that public communication can be significantly enhanced by well-chosen images and data visualization properly sequenced and referenced during a talk.

What 2026 is going to continue to drive home is that AI has a way of exposing the unexamined assumptions in our pedagogy. We’ve been having that conversation about writing for more than two years. For presentations, we’ve barely even started.¹

As I was finishing this piece, I came across a note from Sam Illingworth, who writes the Substack “Slow AI.” In it, he shared a slide deck he generated with Gamma.

His comment: “It took me 10 seconds to generate this slide deck... with the free version & zero additional editing.”²

If that’s where we are now, it’s worth asking whether we’ve figured out what we’re actually trying to teach students when it comes to building and delivering presentations.

How can we ensure our assignments reflect that clarity?

I’m curious how other teachers are handling this. Let me know your thoughts in the comments.

---

Connect With Me

Beyond this newsletter, I work directly with schools, educators, and organizations navigating AI integration. Take a look at my website and reach out - I’d love to hear what you’re working on.

1. Though it might be worth noting how far the conversation has come since the student from Northeastern sued the university and a professor for using AI to help make his presentations.

You can find the Gamma Presentation here: AI Cannot Be Your Friend

2.