For a while now I have worked with a simple rule to decide what the AI does and what a person does in a campaign: AI for informative, objective content; humans for the emotional part.
Why that line
Audiences are fine with an AI summarising specs, comparing prices or organising information. Trust breaks with emotion: when people sense that a message trying to move them was written by a machine, they experience it as fake. And an emotion that feels fake doesn’t just fail: it backfires.
How it splits in practice
- For the AI: product sheets, data-driven ad variants, summaries, performance analysis and first drafts of briefs and keyword lists.
- For the human: the concept, the insight, the claim, the tone, the story and any piece that has to make people feel something.
It’s not nostalgia: it’s performance
I don’t say this out of romanticism. AI saves me hours on what is measurable and repeatable, and I put those hours where they make the difference: understanding the person on the other side. It is the oldest argument in advertising, know your audience, with a new tool to get the mechanical work out of the way.
If a piece has to move people, someone should sign it.