Nib asked, and Nib — askNib's tutor — drew the answer live at a whiteboard. This is the spoken transcript; enable JavaScript to watch it drawn.
Why can't AI count the r's in strawberry? (for a curious adult who uses chatbots but has never studied computing). Cover, in this order: What does an AI actually receive when you type a word?; Where do tokens come from?; Why is strawberry split into pieces the model can't see inside?; How does a model learn spelling at all?; Why does letting it write the letters out fix the count?; What else do tokens explain? — page 5: Spelling Out Fixes Counting
Ask for the count straight away and the model must recall letter facts about hidden chunks. That's memory, and memory here is weak.
Now let it write s, t, r, a, w, b, e, r, r, y. Each letter becomes its own token, sitting separately on the page.
The model reads back the letters it has already written. Counting then means looking at a list in front of it, not recalling facts.
Add a running tally: r one, r two, r three. Each step is tiny, so the answer, three, is simply read off.
So when a chatbot miscounts, ask it to spell the word out first: you're not tricking it, you're giving it letters it can actually see.