Do a practical to back the theory.
Every note in this notebook comes with something you can do yourself. Most need nothing more than the AI tool you already use. Some run right here on this page. They are collected in one place so you can work through them on their own, or alongside the notes they belong to.
Practical 01 — The same half-sentence, five times
Belongs to Note 01: a model doesn't know things, it predicts what comes next.
Open a fresh chat each time, so no earlier answer influences the next. Paste exactly this and nothing else:
Complete this sentence in one line: "The most surprising thing about working in enterprise technology is"
Do it five times. Write down the five continuations side by side. Then, if you like, do it again in a different model. What you are looking at is the point of the note: there was no answer stored anywhere. Each time, the model chose a likely next word, and then the next, and the path diverged.
write down what you noticed. that's the whole practical.
Practicals are added with their notes. Practical 02 is a tokeniser you can type into; Practical 03 is a temperature slider. Whenever a mechanism is better poked than described, there will be something here to poke.
If you run a practical, I would like to know what you found. Results from readers are credited in the notes they sharpen, and in the book. The easiest way to send them is a comment on the LinkedIn announcement of that note.