The companion tool for the AI4LD ebook Introduction to GenAI and LLMs(Unit 2). Step through how a language model actually predicts the next word, then describe a multimodal experience you're building and get a Restrict/Scaffold/Require read on your AI involvement, plus a debrief design grounded in multimodal-analytics research on training debriefs.
At each step, the model isn't just guessing the single most likely next word from the last one: it's weighing the immediate phrase, the sentence built so far, and the broader topic all at once. Click through to see how those layers shift the odds.
3–16 words works best: short enough to step through, long enough to show how context shifts the prediction.
Most of what a learner actually takes away from a multimodal experience gets built in structured reflection after the experience, not during it. Describe your experience and this tool will classify where AI involvement helps or contaminates your evidence, then design the debrief moment itself.
Select everything involved in the experience: check all that apply.
What should learners be able to do, or understand, after this experience?
Who are the learners, and what's their starting point?