What Twenty Years of Advertising Production Taught Me About AI

After two decades directing advertising production — from traditional campaigns to AI-assisted filmmaking — here is what the craft has taught me about where artificial intelligence genuinely helps and where it never will.

I have been making advertising content for a long time. Long enough to have worked with 35mm film, digital conversion, the arrival of non-linear editing, the rise of social media formats, programmatic creative, and now artificial intelligence. Each of these shifts was described, at the time, as either the death of advertising craft or its salvation. None of them were either. They were tools — some more transformative than others, all of them dependent on the same thing: a human being with a clear idea of what they are trying to communicate and why it should matter to another human being.

AI is the most powerful tool I have worked with in my career. It is also the most misunderstood. And the misunderstanding runs in both directions — people who think it replaces everything, and people who think it changes nothing. I want to share what I have actually learned from working with it, because I think the practical truth is more useful than either extreme.

What the Production Floor Actually Teaches You

When you spend years on production floors — managing timelines, arguing with directors of photography about whether a shot works, calming clients who cannot articulate what they do not like about a cut, negotiating between what a brief asks for and what will actually perform — you develop a very specific kind of knowledge. It is not theoretical. It is pattern recognition built from thousands of hours of watching what works and what does not, what clients think they want versus what their audiences actually respond to, and what technical capability enables versus what creative instinct demands.

This kind of knowledge is entirely absent from AI systems. Not because AI is unintelligent — it is extraordinarily capable in its domain — but because that domain is pattern prediction from existing data. It has never felt the peculiar frustration of a shoot running four hours over schedule because the light in a particular location is not doing what anyone anticipated. These experiences are not sentimental background colour. They are the substance of advertising judgment.

The Three Things AI Has Genuinely Changed for Me

First: visual exploration. Before AI image generation tools reached their current capability, exploring visual directions for a campaign was expensive and time-consuming. AI generation tools allow me to show, rather than describe, a visual direction — at a level of polish and completeness that removes the translation problem. We have cut early-stage concept presentation time by approximately 60%.

Second: iteration speed in post-production. AI tools have not replaced the judgment calls in this process. They have reduced the time and cost of executing each iteration, which means we can pursue more refinements within the same budget. The creative quality ceiling has risen because the cost of reaching for it has fallen.

Third: copy and concept variation at scale. AI allows me to generate dozens of coherent variations from a single approved direction, in a fraction of the time. The catch — and this is critical — is that I still read every single one of them. AI copy generation without human editorial review is a mistake.

The Thing AI Cannot Do and Why It Matters

There is a moment in every strong advertising campaign where a decision is made that is not derivable from the brief, not supported by any data, and not obviously correct to anyone in the room. It is a creative risk — a choice to say something in a way that nobody has said it before. AI cannot make these decisions. It is why creative direction — genuine creative leadership — remains irreplaceable in advertising, regardless of how capable AI tools become.

A Practical Note on Client Relationships

Twenty years of production has taught me that clients are paying for outcomes, not processes. They do not care whether a visual was created by a human artist or an AI tool. They care whether it works. My position is that transparency is both ethically correct and commercially sensible. When I use AI tools in a production, I tell the client — because honest relationships produce better work.