For some weird reason, I always had some kind of slight “mental hesitation” with the meaning of data encoding versus decoding. Which one goes in what direction? To be honest, I have the same kind of weirdness with other concepts: daylight saving time for instance (are we gaining or losing an hour? I can never tell, for many days after a change).
So I wanted to create a diagram to illustrate the dichotomy between encoding and decoding, for a course I’m creating on software engineering. So one way to “create with AI” would be to ask one: “Can you please create a diagram to illustrate the difference between data encoding and decoding”.
But I know in advance, the kind of diagram I will get, from Gemini (Nano Banana). So I knew that in this particular case, I want to do it myself, with Excalidraw (which I love greatly). So I asked: “Can you describe a visual schema that would explain it”.
Don’t do it yourself, just describe it to me. Then I got my answer, something like this:
Source Domain Target Domain
(meaningful to humans) (useful for machines/systems)
[ Data A ] ──encode──▶ [ Data B ]
▲ │
│ ▼
decode ◀─────────────────────┘
Or something like this:
meaning → (encode) → representation → (decode) → meaning
Then I thought ok, I can do that. Then I asked myself, to begin how can I represent meaning, the real world? A cloud, of course. A cloud is clearly the real world.

Then I wanted to represent the computer’s side of it.. I wasn’t sure.. a box? Something else? I asked the AI, got a few suggestions, and went with the obvious: a schematic computer (schematic enough that I can put a word label in it, like I did with the cloud).
And then I had it:

So what I mean by this is that the meaning of “creating with AI” does not have to be fully polarized between “the AI did it entirely by itself” and “no I kept it fully real, no AI”. Creating with AI can be a process, where you use the AI to think, brainstorm and iterate, in order to create something, together.