Demystifying large language models

• Audience any audience, up to a few hundred
• Duration 15 minutes
• Format sheets

TED-style: the speaking carries it and the slides are cues, not visual aids. The audience is handed search sheets as the house lights come up, runs the model for four minutes by show of hands, and then learns what it just did. The three source books are never named until near the end, and nothing on the printed sheets names them either, so a hall that has just generated “the old man sat in the hat” finds out it was Hemingway, Poe and Dr Seuss—in that order, because Seuss is the biggest recognition hit and has to land last.

We gave it at the ANU Visionaries Among Us Showcase in August 2026, to a hall of about 360. The deck below is that talk with the event-specific slides taken out; the deck as delivered is linked alongside it.

Running order#

Target thirteen minutes against a fifteen-minute slot; the run is where the spare time goes.

Only three slides carry real text: the algorithm, the role mapping and the three-book reveal. If a slide is tempting you to read it out, cut it rather than the sentence you were going to say.

You will need#

• one search sheet per seat, dealt from the three texts together, with the title and author left off every sheet (the CLI’s sheets subcommand takes --title and --author overrides); the deck’s comment block has the exact recipe, and the deal is seeded so it’s reproducible
• ushers to hand the sheets out on cue, fast: “take one, pass the rest along”
• a whiteboard, or the deck’s annotation layer, to write the growing sentence where the back of the hall can read it
• a projector for the slides

Before you deliver#

• rehearse the hand-out: the sheets go out on slide 4, not before, and the run can’t start until nearly everyone is holding one
• pick loud seeds in case you need to restart; the deck notes list them for each text
• write big: fifteen tokens at whiteboard size is the whole output, and the back row has to be able to read it
• the as-delivered deck has ANU-at-80 slides near the end that were for that event; the deck above already leaves them out

Using the slides

• Demystifying large language models (15 min)
• as delivered at the ANU Visionaries Among Us Showcase, 15 Aug 2026

The slides run in your browser, and every slide carries full presenter notes—what to say, what to watch for, and how long each part should take. A few keys worth knowing:

• f toggles fullscreen—do this before you start presenting
• s opens the speaker view in a separate window, with the presenter notes, a timer, and a preview of the next slide (drag it to your laptop screen and put the slides on the projector)
• o shows an overview of all slides, handy for skipping ahead if you're running short on time