AI can get us to plausible incredibly fast.

But plausible isn’t the same as right.

Superside has a framing I like: AI-generated creative can often get teams “90% there.”

Not because 90% is some universal benchmark, but because it captures the tension well.

Judgment has to run through the whole process.

Before generation: What problem are we solving? Who is it for? What does “good” look like?

During generation: Which direction has potential? What feels generic? What’s interesting but wrong?

After generation: What needs refinement? Does it feel like the brand? Would I put my name on this?

Here’s the detail that stuck with me: in Superside’s own process, strategic planning accounts for roughly 80% of the time.

And that tracks with what I’ve seen using AI creatively. Generating more options usually isn’t the hard part anymore.

The harder part is knowing which direction is actually worth pursuing, what needs another round, and what simply isn’t good enough yet.

If a team can generate 50 options in an afternoon, the number of options isn’t especially valuable.

Knowing which two deserve more work is.

Taste. Context. Restraint. Experience. Knowing when something looks impressive but is still wrong for the problem.

The goal isn’t to protect creative judgment from AI.

It’s to build AI into a process where judgment still decides what deserves to ship.

AI can accelerate the work.

It doesn’t get the final say.

Reference: Superside, “How to Achieve AI Creative Quality Control That Goes Beyond Speed.”