One of the first things people do with generative AI is ask for more.
Ten names. Twenty headlines. Fifty logo concepts. A hundred campaign ideas.
Why not? The marginal cost of another option is almost zero.
For decades, creative work was partly constrained by the cost of making the options. AI has weakened that constraint.
What it has not weakened is the cost of deciding between them.
More choice can hide weak thinking
If a team cannot agree on the problem, generating 100 answers does not help.
It gives the disagreement more objects to attach itself to.
Now the conversation becomes: “I like 17.” “Can we combine 17 with the colour from 43 and the icon from 62?” “Let’s show all 100 to the founder.”
This is how a creative process turns into a buffet.
Quantity feels democratic. Good design usually requires somebody to edit.
The job is convergence
A studio does not add value by forwarding every possibility the machine produced.
It adds value by reducing possibilities intelligently.
What fits the brief? What is legally risky? What looks too close to the category leader? What creates production problems? What requires an explanation every time someone sees it? What works beautifully in the presentation but badly in the actual business?
A shortlist is an argument
When we present three directions instead of thirty, that does not mean only three things were considered.
It means the work of elimination happened before the meeting.
A good shortlist should say: these are the most credible territories; here is why each one answers the problem differently; here is what we would recommend.
This is much more valuable than making the client perform the designer’s editing job.
AI makes criteria more important
Before generating options, define what a good option has to do.
For a name: easy to pronounce, ownable, not overly descriptive, usable across markets, no unfortunate linguistic associations.
For a website hero: explain the business quickly, establish the right tone, support a clear action, work with available imagery.
For a visual identity: distinctive in category, flexible across applications, reproducible within budget, recognisable at small sizes.
Now you have a filter.
Without criteria, every option becomes a matter of taste in the weakest sense: I like this one.
The paradox of AI creativity
The easier it becomes to create variations, the more valuable restraint becomes.
The skill is no longer proving that we can think of another answer.
Of course we can.
The skill is recognising when another answer is not improving the decision.
AI can multiply possibilities. Creative direction is knowing when to stop.