Turn Complex Research into Classroom-Ready Presentations with AI
In Short
Turn complex research into engaging, classroom-ready presentations with AI. Discover how AI can simplify research, organize key insights, and support effective teaching.
Share :
Turn Complex Research into Classroom-Ready Presentations with AI
Tuesday, half past six, a government degree college on the Nalgonda road. Sridevi (a composite) has a 30-page paper on groundwater depletion open on her laptop and a printout of the same paper on the desk, the margins of pages 14 to 19 already dark with pencil. On Thursday she has to teach it to a second-year BSc class that did one semester of hydrology, knows what an aquifer is, and has never read a journal article in its life. The paper has six figures, two tables, a trend test she will have to explain without the statistics course the students have not yet taken, and a water-balance equation with eleven terms.
Two floors up, a research scholar is doing a near-identical job for a departmental seminar on Friday, and in an engineering college outside Vijayawada a final-year student has been handed a machine-learning paper for the “paper presentation” that counts towards internal marks. Three different rooms, one task: take a document written for forty specialists and make it land with forty people who are not.
I have done this job badly for years, mostly by pasting the abstract on slide two and apologising for it. I stopped two years ago. The current routine uses an AI slide generator for about twenty minutes and my own pencil for about two hours.
Three ideas, on the back of the printout
Before any software opens, write down the three things the class must still remember on Saturday. Not the findings, the ideas. For the groundwater paper, Sridevi’s three were: water tables in hard-rock districts have been falling for most of two decades; a trend can be measured from messy well data if you choose the right test; and the authors stop short of saying what should be done, on purpose.
Three is the limit. A paper carries thirty ideas and a lecture carries three, the rest is for the students who come to your room afterwards.
If you cannot write the three, you have not finished reading the paper, and no tool fixes that.
The prompt does most of the work the slides will get credit for
The paper is invented for this column, a composite of several real ones, so nobody should go searching for it: call it a 2024 study of pre-monsoon water levels in 212 observation wells across six southern Telangana districts, 2005 to 2022, checked against satellite gravity data. Sridevi attaches the PDF to Quillbot’s ai presentation maker through the Attach Files button, which takes PDF, Word, PowerPoint and image files, sets the slide count to ten, and writes a prompt that is really a lesson plan:
Attached is a 30-page research paper on groundwater depletion in six districts of southern Telangana. Build a lecture deck for second-year BSc students who have done one semester of hydrology, know what an aquifer is, and have never read a journal paper. Plain English, short sentences, no abbreviation left undefined. Slide 2 must be a “terms we need” slide that defines piezometer, pre-monsoon level, hard-rock aquifer, command area and trend test in one line each. Then cover: why the study was done, where and how the data was collected, the two or three main findings, and what the authors say they could not conclude. Keep every number exactly as the paper states it. Ten slides.
The class level is the sentence to get right. Leave it out and the generator writes for the paper’s own readers, which is to say it rewrites the abstract ten times. The defined-terms instruction matters almost as much, because a second-year who loses “piezometric” on slide three stops listening on slide four.
Generation does not take long. What comes back is a set of slide titles with speaker-ready text under each, and then a choice of layouts and themes. Pick the plainest theme. A lecture deck on a projector that has not been cleaned since 2019 does not need gradients.
One abstract, four lines
The abstract is where the editing begins, because the generator will try to keep it and you must not let it. Here is the composite paper’s abstract, more or less as such abstracts read:
This study integrates piezometric observations from 212 monitoring wells (2005 to 2022) with terrestrial water storage anomalies to characterise spatio-temporal trends in groundwater depletion across six hard-rock districts of southern Telangana. A non-parametric trend test and slope estimator were applied to pre-monsoon water-level series; results indicate statistically significant declining trends in 68 per cent of wells, with a median decline of 0.31 m per year, attenuated in canal command areas after 2019. Findings underscore the need for aquifer-specific abstraction regulation.
And here is the slide that replaced it, after the generator’s first attempt and Sridevi’s second:
Where the water went, 2005 to 2022 Pre-monsoon water levels fell in about two of every three wells the study measured. Typical fall: roughly a third of a metre a year, so a bore-well that reached water in 2005 needs about five more metres of pipe today. Areas fed by canals did better after 2019. “Regulation is needed” is the authors’ suggestion, not their finding. Keep the two apart.
The generator wrote none of the last line. The five-metres line it did not write either; that came from Sridevi doing the multiplication on the printout, and it is the line the class will repeat at home.
Keep, compress or drop
Every section of a paper has a different fate in a lecture. This is the table Sridevi pencilled on the inside cover before she began deleting slides:
Paper section | Keep, compress or drop | Why |
Introduction and literature review | Compress to one slide | The class needs the one gap the authors found, not twelve years of citations |
Study area and data | Keep | Six district names and a map of wells make it real for students who have watched those bore-wells go dry |
Methods | Keep two figures, drop four | Two figures a class can read beat six it cannot |
Discussion and limitations | Keep, close to verbatim | The authors’ caveats are the lesson in how research talks about itself |
The scholar’s seminar deck follows the same table with one change: methods get three slides, not three lines, because the room is full of people who will ask about them.
The hard edit, which is most of the two hours
The generated deck is a draft by a diligent assistant who has not read the paper the way you have. The editing happens in the tool itself: section titles and text are editable, sections can be reordered, deleted or added, and images can be cropped.
Strip the jargon the prompt did not catch. “Spatio-temporal” becomes “across the six districts and over the years”. “Attenuated” becomes “slowed down”. “Statistically significant” stays, with one plain line under it, because the students will meet it again in the third-year paper and might as well meet it here with a friend.
Check every claim against the paper, with the page open. Every number, every “most”, every “sharply”. The generator moved “68 per cent of wells” to “most wells” on one slide, which is true, and to “a majority of districts” on another, which the paper never says, because the result is about wells and not districts. That one took four minutes to catch and would have taken four seconds to repeat in an exam answer.
Keep the authors’ caveats. The limitations paragraph on page 24 said the well network was thin in two of the six districts and that the satellite data cannot see below district scale. The first generated draft dropped both. They went back as a slide of their own titled “What the authors say they do not know”, and that slide, in my experience, is what the sharpest student in the room writes down.
Replace the generated images with the paper’s own figure. The tool can generate charts and images to match a topic, and for a sales deck that is useful; for a lecture on someone’s measured data it is a trap, because a generated chart of groundwater decline is a picture of nothing. Sridevi took a screenshot of Figure 3 from the PDF (a photograph of the printout works too, cropped square in the editor), uploaded it as a PNG, and put a credit line under it: Figure 3, from the paper, authors and year. The map of wells from page 7 followed, cropped from the PDF and credited.
Add a “pause and ask” slide. One slide, one question, nothing else on it: “Your village’s bore-well: deeper or shallower than when you were ten? Ask at home and bring the answer on Monday.” It goes straight after the results slide, and it does more for retention than any animation, which the tool does not provide anyway.
She also cut the generator’s closing “Thank you” slide. Nobody learns anything from one.
Which file goes on the pen drive?
The department laptop runs an old copy of Office, is shared by eleven lecturers, and has a projector cable that must be held at an angle. Quillbot’s ppt maker page is the door to use when the end product has to be a PowerPoint file, because the download is a PPTX that opens there with the fonts where you left them, and the same generator sits behind both pages. Export the PDF too and keep both on the pen drive; the PDF is what you show when PowerPoint decides Thursday is the day to update itself.
The tool does not add speaker notes, transitions or animations. For a lecture that is roughly the right set of absences. Write your notes in the PowerPoint notes pane after export, three lines a slide, and leave the transitions alone. If your college has moved to Google Slides, there is no native export yet; upload the PPTX and it opens.
The scholar shared her draft deck with her guide by link the night before the seminar, and he changed two slide titles while she watched, which is the sort of thing that used to take a day of email attachments.
Where it falls short
Equations and diagrams. The water-balance equation arrived as a line of text with its subscripts flattened, and the conceptual cross-section of the aquifer came out as generic art with the wrong number of layers. Equations go on the whiteboard, where you can build them one term at a time anyway, and diagrams come from the paper, cropped and credited.
Citations are yours to add. The deck carries no reference list. Put the full reference on the title slide and again on the last, and if you mention work the authors cite, type those citations yourself from the paper’s own list. Quillbot’s Citation Generator formats them; it does not find them.
Regional-language classes. A good part of the degree colleges in Telangana and Andhra teach in Telugu or in a mix, and the slides come out in English because the paper is in English. Sridevi translates the four plain lines on each slide herself and keeps the English term in brackets, since the English term is what the examination will use. Which Telugu word to use for “aquifer” in a classroom is a teaching decision, and I would not hand it to any software.
The product page still carries a Beta label. Treat it as one: if a section comes back thinner than you asked for, rewrite it by hand in the editor, which is faster than arguing with a prompt.
Questions from the staff room and the hostel
Is presenting a paper this way a form of plagiarism? Presenting someone’s published work with full credit is what a seminar is. The problem arises when the generated summary stands in for your reading, so that you are presenting the tool’s understanding rather than yours. The three-ideas step and the page-by-page check are what keep it your presentation.
Can I attach a whole thesis chapter? The tool accepts PDF and Word files, so yes. An 80-page chapter still gets three ideas and ten slides; if it genuinely has six ideas, make two decks and present them a week apart.
Will it write my speaker notes? No. Nor transitions or animations. The notes pane in PowerPoint is where those go, after export.
Does it cost anything? The presentation maker is free to use. A Premium plan adds unlimited presentations, faster generation and access to the rest of Quillbot’s tools, which matters for a scholar presenting every fortnight and not at all for a lecturer doing this twice a semester.
Tuesday night, for Thursday
Pick the paper. Read it with a pencil, and write the three ideas on the back page before you open anything. Then attach the PDF, write the prompt with the class level in the first sentence, set the slide count to ten, and give the generator its two minutes. Spend the rest of the evening with the printout open beside the draft, striking out one word on every slide and putting the authors’ caveats back in. Screenshot the figure you want them to remember, credit it, and export the PPTX before you sleep. Thursday will still be hard, but it will be hard for the right reason, which is the paper, and not the slides.
