The thing that changed first was not the output. It was the silence before the work starts.
There used to be a specific kind of creative friction at the start of a brief. You had an idea forming, and then two or three days where you were pulling references, building a mood board, looking for that one image that crystallised the feeling. That friction was uncomfortable and, I think, useful. It was where the idea sharpened.
Now that phase compresses to a few hours. Sometimes less. The friction is still there, but it is shorter. I am not sure yet whether that is a gain or a loss. Probably both.
The brief-to-concept phase
Before AI tools, a concepting session with a new brief looked like this: a day of desk research, a day of pulling images from Pinterest, Behance, Google Images, photography archives. A day of organising those references into something that communicated a direction to the team. Three days before we were anywhere near a concept.
Now I can run a prompt pass in Midjourney and have a hundred visual directions in under an hour. Flux gives me something closer to photographic quality when I need it. I can describe a feeling, a colour relationship, a spatial mood, and see something back in minutes.
The speed is real. But here is what that speed reveals: most of what the tools generate is average. It is the mean of everything they have seen. To find something that actually fits a specific brand, a specific audience, a specific moment, you need taste. You need to know what you are looking for before you see it. That is still the director's job, and it has not got easier just because the pool of material arrived faster.
What has changed is that I now arrive at a concept presentation with more options, more visual texture, and a clearer picture of the directions I am ruling out. That is genuinely useful for clients and for the team.
3D and asset generation in immersive work
This is where the change has been most concrete for me as a CD working in the XR and immersive space.
Generating a 3D asset from a text prompt or a reference image used to take a specialist days. Meshy and Luma AI have compressed that to minutes for certain asset types. Runway has changed how I think about motion and video concepting. In a production context, this matters: we can iterate on a look four or five times in the time it would have previously taken to iterate once.
For brand activations, this changes the economics of the early creative phase. I can show a client what a 3D environment will feel like before we commit to building it. That is a better conversation.
But consistency is the catch. AI-generated 3D assets often drift from a brand's established visual language in ways that are hard to articulate but immediately visible. The colour science is off. The material quality feels generic. A product looks close, but not quite right. Getting from "approximately right" to "brand-accurate" still requires human oversight and, often, significant manual refinement. The tools accelerate iteration. They do not replace the art direction required to land on something consistent.
I have also found that generative pipelines are most useful when they are tightly constrained. A vague prompt produces a vague result. The more specific the brief, the more specific the output. That sounds obvious, but it means the skills that make a good CD, writing precise briefs, thinking in specifics rather than generalities, making editorial decisions quickly, those skills matter more in an AI-assisted workflow, not less.
"AI raises the floor. The average piece of creative work is better than it was two years ago. But I have not seen AI raise the ceiling. The most memorable campaigns I have encountered recently all have a specific point of view that no prompt produced."
What AI still cannot do in immersive work
Spatial reasoning. That is the short answer.
When I am designing an experience for a 10-metre by 10-metre footprint, I am thinking about sightlines, about where someone's attention will be three seconds after they walk in, about how the experience changes when there are forty people in the space compared to four. I am thinking about sound bleed, about the physical arc of a visitor's journey, about what the experience communicates from across the room before anyone has engaged with it directly.
None of the current tools reason about physical space in that way. They generate content, not experiences. The gap between content and experience is where creative direction lives in immersive work.
The other thing AI consistently misses is the unspoken brief. A client brief says one thing and means another. The explicit ask is often a proxy for a business problem or a brand anxiety that has not been named directly. Reading that, asking the right questions, repositioning what the work needs to do before a single concept is generated: that is one of the most important parts of the job. I have not found a tool that does it.
There is also creative risk. AI optimises toward what is plausible given its training. The decisions that make a campaign genuinely unexpected, the choices that feel slightly wrong until they are right, those require someone willing to make a call and defend it. Tools do not take creative risks. Directors do.
AI raises the floor, not the ceiling
This is the observation I keep coming back to. The average quality of creative work has gone up. Someone with access to the right tools and a reasonably good eye can produce something that looks polished, considered, and professional. That is a meaningful shift.
But I have not seen AI raise the ceiling. The work that actually cuts through, that people remember, that becomes a reference point for the industry, still comes from a specific and committed creative perspective. It comes from someone who decided what they wanted to say before they thought about how to say it.
What AI has done is make the distance between a good idea and a mediocre execution much smaller. That is useful. It is also what makes genuine creative direction more valuable, not less. When execution is cheap, the idea and the judgment behind it are what determines the outcome.
How this changes the brief
Clients can now see concepts faster. Sometimes they see AI-generated visuals before we have had a proper conversation about strategy. That creates a specific problem: a visual that looks resolved can foreclose creative thinking before it has started. I have had to be deliberate about sequencing. The concept conversation has to happen before the visual conversation, or you end up refining the wrong thing.
The harder conversation is about fees. Some clients see AI-generated imagery and conclude that the creative process is now cheaper because a tool did part of the work. I understand why that logic seems appealing. But it confuses the cost of output with the cost of direction.
The time saved in asset generation does not come from the creative director's fee. It comes from production. The creative direction, the strategy, the judgment about which direction to pursue and why, the stewardship of a campaign from brief to delivery: that work is unchanged. In some respects it is harder, because more options means more decisions, and making good decisions quickly is the core skill.
What AI changes for clients is access to better-looking early concepts at a lower production cost. What it does not change is what makes those concepts worth pursuing. That still requires a person who knows what they are doing.
Key observation
The tools have changed what the first week of a project looks like. They have not changed what makes the project worth doing. A brief still needs someone to read it, question it, and decide what the work is actually for. That part has no shortcut.
I am still figuring out where this lands. My workflow in March 2026 looks different from my workflow in 2023, and it will probably look different again by the end of this year. What I am more confident about is the underlying principle: the tools serve the direction, not the other way around. The moment I catch myself choosing a creative direction because the AI output looked impressive, rather than because it is right for the brief, I have made a mistake. The discipline is the same as it always was.
Common questions
How has AI changed the role of a creative director?
AI has compressed the early phases of creative work: reference gathering, mood boarding, and initial concept exploration now happen in hours rather than days. But the role of a creative director has not shrunk. If anything, it has become more critical. With AI generating more raw material faster, the director's job is to filter it, shape it toward a specific brand position, and make the calls that give a campaign its point of view. The taste and judgment required for that has not been automated.
Can AI replace a creative director on an immersive project?
No. Immersive work involves spatial reasoning, physical constraints, and an understanding of how an audience moves through and responds to an environment. Those decisions require human judgment: how a 10-metre installation reads from across a room, where a visitor's attention will land first, how the experience needs to feel 90 seconds in. AI can generate assets and accelerate iteration, but it cannot hold or reason about the full spatial and experiential context of a brief.
What AI tools do you use in your creative workflow?
I use a mix depending on the phase. For early concepting and mood boarding: Midjourney and Flux for image generation, ChatGPT for rapid concept framing. For 3D and asset work: Meshy and Luma AI for generative geometry, Runway for motion and video. For generative and interactive experiences: custom pipelines built on top of open-source diffusion models. None of these tools are used in isolation. Each one requires a brief, a point of view, and a director to make sense of the output.
Does using AI mean creative projects cost less?
AI compresses certain parts of production: asset iteration cycles, initial concept exploration, some 3D geometry tasks. But it does not replace creative direction, strategy, or the expertise required to make a campaign distinctive and on-brand. Clients who expect a significant fee reduction because AI generated an image are misunderstanding where the value sits. The cost of creative direction is the judgment, taste, and experience that determines which direction is right for a specific brand and audience. That part is unchanged.