When AI can make the first draft, what do creatives sell?

A prompt is not a brief

Imagine a local theater needs a poster. “Make it cinematic” is a prompt. The brief also says who the show is for, what it is actually about, which images the theater may use, where the poster will appear, what must be readable on a phone, and who approves it.

A generator might return twenty striking options. If they all sell a disaster film while the show is intimate and funny, that speed was a detour. Somebody has to notice the mismatch, explain it, and choose the direction that fits the work.

Adobe Firefly can turn a text prompt or an image into video, and Canva’s Magic Studio puts AI tools across a design workflow. They do not tell us whether the output answers a particular brief. Could we use the result for a real person, in a real business, with someone responsible for what goes live?

The price of a first draft can fall. The job still has stages.

If the service is only “I will press Generate for you,” a client can reasonably ask why they should pay for it. A finished piece may still require discovery, concept choices, truthful claims, consent and rights checks, mobile crops, accessibility, revisions and delivery in the formats that will actually be used. Not every job needs every stage.

Show the rejected options and why they failed. Show what changed after the brief. State the deliverables. Do not charge a mysterious premium because a process sounds “human”; make the useful difference visible.

What about the craft?

For someone who spent years learning to draw, film or edit, a new tool can feel as though it is competing with a piece of their identity. Telling that person to “adapt” is not a plan. Nor does a cheaper first draft guarantee a stable market for everyone who made first drafts before.

The practical question is narrower: what does your craft let you see and decide that the tool does not settle for this project? It may be the physical skill. It may be the ability to direct a scene, keep a campaign coherent, spot a false note, or challenge an instruction that would mislead the audience. Some buyers will choose the cheapest acceptable asset. Others need someone who can account for the result. Neither outcome should be assumed without testing the actual offer.

Rights belong in the workflow

In the United States, the Copyright Office’s report on AI and copyrightability says AI assistance does not by itself rule out protection, while the human contribution to expressive elements matters. Based on current tools, the report concludes that prompts alone do not give a person enough control over the output to make them its author. The answer for a particular work depends on its facts, so the practical move is to document what people selected, arranged, rewrote and made.

That is not the only rights question. A working creative brief should record source assets, likeness permissions, client materials and where the finished piece may be used. A convincing image is not automatically cleared for a campaign.

Test the whole job

1. Write the brief first. Name the audience, promise, facts, placement, constraints, rights and decision maker.
2. Use AI for exploration where it helps. Try rough layouts, alternative framings or draft language without treating the first result as a finished asset.
3. Put a person in charge of release. Check accuracy, originality, consent, mobile presentation and final selection.
4. Measure the full workflow. Count prompting, filtering, repairs, licensing, approvals and delivery. Compare like with like.

If the tool saves time on the whole job and the work improves, keep it. If it moves the labor into cleanup, change the process. If it makes this task worse, use another method. The result matters more than the demo.

The future question for a creative business is not whether a machine can produce a first version. It can. The harder question is who can say, with evidence, “This is what we meant, and this is ready.”