40 Questions Creative Directors Should Ask Before Trusting AI
Your taste and judgement are what separate your team's work from what any brand can generate in seconds. These questions help you keep your creative standards intact when AI makes everything look competent at first glance.
These are suggestions. Use the ones that fit your situation.
1If I showed this Midjourney output to a creative from ten years ago, what would they say we gave up to make it faster?
2Am I evaluating this work against what the brand needs or against how easy it was to produce?
3Can I articulate why this Adobe Firefly visual is better than three other versions, or does it just feel finished?
4What creative decision did the AI make that I would not have made, and is that difference a problem or a discovery?
5If my team had produced this instead of Midjourney, what feedback would I give them?
6Does this DALL-E result look distinctive for this brand or just competent in general?
7What does this output tell me about my team's current thinking if they're using it as reference?
8Am I comfortable presenting this to the client knowing they will see the same result if they prompt ChatGPT themselves?
9Which part of this work required human creative judgement and which part could have come from any tool?
10If I strip away the speed advantage, is this work better than what we made six months ago?
On Briefing and Thinking
11Have we written a brief that AI could execute or a brief that humans need to interpret?
12When I used ChatGPT to develop the brief, what thinking did I skip that a traditional brief would have required?
13Can my team see the actual brand problem in this brief or just the AI-generated tone?
14Did using AI to write the brief make it easier to produce work or harder to judge it?
15What single line in this brief would disappear if I removed all the ChatGPT language?
16Is the brief specific enough that it would reject 80 percent of what Midjourney could generate, or just 20 percent?
17Did I brief my team to solve a problem or to curate Runway ML outputs?
18What strategic choices are hidden inside the AI-generated brief that my team needs to understand but probably won't?
19If we didn't have these AI tools available, would this brief be stronger or weaker?
20Can anyone on my team recite the core creative idea without reading the AI language?
On Team Instinct and Risk
21Is my team developing their eye for exceptional work or learning to spot outputs that Midjourney generates well?
22When someone on the team suggests an idea that seems risky, am I more likely to say yes or to ask for an AI version first?
23What have we stopped trying because AI is competent at the safe version?
24Are we hiring for creative instinct or for the ability to brief AI tools effectively?
25When did I last push back on a ChatGPT output and ask for something weirder?
26Is my team finding unexpected solutions or hunting for prompts that match what they already imagined?
27What creative move would I have rejected last year that I'm now accepting because AI made it easier?
28Have we created space in the process for intuition or are we moving too fast to trust it?
29When I review work now, am I noticing more about what the AI can do or what the brand needs?
30What would happen if I told my team to develop three ideas and none of them could use AI tools?
On Brand Identity and Differentiation
31Could a competitor generate visually identical work by prompting Midjourney with the same brand guidelines?
32What about this work would be impossible to replicate by someone outside the agency asking DALL-E the right questions?
33Is the brand's distinctive voice coming from our thinking or from how we're instructing ChatGPT?
34If I remove the AI-generated elements, what remains that is specifically this brand and not just this template?
35Does using Runway ML for these shots mean we're committing to that aesthetic in ways we cannot undo?
36When the client sees this work, will they feel it's distinctive or will they wonder if they could get something similar faster?
37What creative characteristic of this brand is becoming harder to maintain because AI defaults to something else?
38Are we building brand identity or are we building the ability to generate on-brand outputs at speed?
39If every Firefly output for every brand in our portfolio came from the same tool, what makes each one specific?
40What part of this brand's identity would be lost if we had to describe it in prompts instead of showing it in work?
How to use these questions
Before sending AI output to the team as reference, ask yourself whether it narrows their thinking or expands it. If it narrows, brief them differently.
When evaluating whether something is good work, cover the tool name and judge what remains. That remainder is your real standard.
Set a rule: at least one major idea per project must come from a conversation, a sketch, or a risk that AI would have optimised away.
Your job is not to become good at prompting. Your job is to know when good work comes from thinking and when it comes from asking better questions of a tool.
Watch your team. When they stop proposing ideas and start proposing prompts, you have lost something important. Name it directly and change the process.