The Data First
Let me start with what's actually being measured. Researchers at several institutions have run blind taste tests comparing AI-generated creative outputs — images, short musical compositions, brand copy, short video clips — against human-made equivalents. In multiple categories, participants preferred the AI output without knowing the source.
This is real. I'm not going to minimize it. When I first read the methodology on these studies, I looked for the flaw — the rigged sample, the cherry-picked outputs, the poorly matched human comparators. Some studies had weaknesses. But the pattern across multiple independent experiments is hard to dismiss: AI is producing creative work that general audiences find compelling, sometimes more compelling than human-made alternatives.
My honest reaction isn't panic. It isn't dismissal. It's a more uncomfortable feeling: recognition that the conversation I need to have with myself about what I actually do well is overdue.
What "Surpassing" Actually Means
There's a critical distinction the headlines miss. When AI outperforms humans in creative tasks in blind tests, it's outperforming average human output. That's not a minor caveat — it's the entire story.
The AI is compared against a pool of humans. That pool includes everyone from weekend hobbyists to working professionals. The average human creative output is, by definition, average. AI trained on millions of examples has learned to produce the statistical optimum of "what people find appealing" — which looks a lot like the average of what professionals have been doing for decades.
What AI cannot do — yet, possibly ever — is produce work that exceeds the upper range of human creative output. The most striking photographs, the most resonant films, the most unexpected music don't succeed because they optimize for average preference. They succeed because they break something — convention, expectation, comfort. Breaking things requires understanding the rules well enough to violate them with intention. AI follows patterns. Breaking patterns requires knowing why you're breaking them.
Where It Genuinely Challenges Me
I want to be honest here, because the comfortable response is to dismiss AI creative capability and move on. But honesty requires admitting where the challenge is real.
For certain categories of commercial creative work — brand imagery, product video, social media content — AI is already competitive with mid-tier professional output. Not exceptional output. But the output that represents 80% of the market by volume.
That 80% is where a significant portion of working creatives earn their living. Not because they're mediocre — but because most client projects don't require exceptional. They require reliable, on-brief, on-budget, on-time. AI delivers all of that for a fraction of the cost.
If my value proposition is "I produce reliable, on-brief commercial content," I have a serious problem. And the honest truth is that I spent the first five years of my career in exactly that value proposition.
The Recalibration
What AI's creative performance benchmark has forced me to do is clarify where I actually create value that AI cannot replicate. I've had to get more specific than "creativity" — because that's no longer a differentiator. I've had to identify the specific types of creative decisions that require what I actually have and AI doesn't:
- Client-specific context: Understanding a brand's history, internal politics, what the CMO is trying to prove to the board, what failed last year and why.
- On-set leadership: Reading talent, adjusting direction in real time, knowing when to push and when to let a moment breathe.
- Narrative structure across a campaign: Not just making one good piece, but making 12 pieces that build a coherent brand story over time.
- Taste that knows when to break rules: Intentional violation of visual convention in service of a specific emotional effect. This requires understanding why the rule exists — something AI doesn't have.
What I've Changed in My Practice
Knowing that AI can produce average creative work faster and cheaper than I can, I've made some concrete changes:
First, I've stopped competing on volume. I don't pitch "content at scale" anymore. Brands who want content at scale should use AI — it's the right tool for that job. I pitch creative direction, campaign strategy, and elevated production.
Second, I've started using AI as a collaborator rather than a competitor. I use AI to generate the "average" outputs that I then elevate — it handles the first draft, the reference frames, the rough options. I handle the final creative judgment.
Third, I've invested more in the parts of my practice AI can't touch. Client relationships. Understanding brand strategy deeply enough to challenge briefs. Directing talent in ways that produce genuine emotion.
The Honest Conclusion
AI is outperforming average human creative output. That's real and it matters. But "average" was always a vulnerable position. The right response isn't to defend average — it's to stop being average. That's a message the creative industry has needed to hear for a long time. AI is just delivering it faster than anyone expected.
