What Matters and What's Filler

When it's this easy for AI to fill a blank page, how do you decide which parts it should fill and which parts you need to decide yourself? Any hole you leave, it will fill. The challenge is knowing which ones it should. Otherwise you end up with busy work that has no immediate impact.

Whenever you produce something, some parts really matter: they make the difference between it being good and bad. Others are filler. They need to be filled in, but it won't matter much whether you choose A or B.

Take a portrait. What matters: the face looks like the person, the height and body type match, perhaps the style of clothing is representative. What's filler: the colour of the background, which earrings, the colour of the shirt.

If you don't separate the two, AI fills in everything, the important decisions and the filler alike. And the person receiving the output can't tell which is which. Getting the shape of the face right looks just as deliberate as the choice of earrings, which the model picked at random.

That leads to two habits.

When you prompt, make clear which elements are important. A decent prompt includes this, and generally not much else. Then make clear how much filling you want. Is it just your key points in a more structured way, or should it add material? If so, what, and how much?

When you send AI output to someone else, tell them which parts are filler. It gives them room to follow that part or not.

It's also the first thing I check when I step in as a fractional CPO: which parts of the plan were decided, and which were just filled in.