
The Recognition Recession: Human Edits, AI Clips, and the Future of Video
Remember when CGI had that telltale digital sheen? When AI writing could be spotted by its use of ‘fostering’? When generated images had obvious distortions around hands and faces?
When the Will Smith eating spaghetti video dropped in March 2023, it was easy to see the distorted and surreal depiction was clearly AI, but our feeds are filling faster than ever with content created that makes us pause and you find yourself questioning... Is this real?
We're experiencing a "recognition recession" - a steady decline of our ability to name artificial content. The earlier comfortable boundaries have vanished, leaving us with a lack of confidence in our own perception.
Back in March, Puma partnered with agency Monks to create a fully AI-generated ad in five weeks, using hundreds of AI agents working collaboratively from concept to completion. When Puma’s technically proficient ad was tested with audiences, it resonated, but something was off. It felt flat, a bit too obvious. It lacked something. The human touch.
Human creativity is like a fingerprint - it goes deeper than the technical execution we’re seeing. The choices that don’t make obvious sense, the imperfections that feel intentional, the surprises that catch out even the creator. The vulnerability of personal experience, the pause between thoughts. Where AI deals in pattern recognition and analyses millions of examples to create something new, it doesn’t quite get the nuances around human experience. The leap that defies logic, the happy accident that becomes the most memorable moment. Not yet anyway.
Recently, PJ Accetturo created an ad for Kalshi in just two days using AI tools, ideating with Gemini and ChatGPT for a rough script, then Google VEO 3 to generate 300-400 iterations to get 15 usable clips that edited together is currently at 30M+ views. The fact that a single filmmaker can produce broadcast-quality content in two days with 95% less production costs than a standard film crew opens up endless possibilities with concerns on either side. What does this mean for video content moving forward?
Sure, the low cost of creating content is a huge advantage, but the conversion rate from iterations to usable clips sits somewhere between just 3.75% and 5%. That kind of ratio would never fly on a film set. Yes, there are legendary examples of endless retakes, like Jackie Chan’s Dragon Lord in 1982, where he reportedly shot one stunt sequence nearly 3,000 times. But that was a meticulously choreographed scene, not a rough eight-second clip. And while it's impressive that AI can churn out 300 clips in two days, there's still a human at the helm, crafting the prompts, shaping the ideas, deciding what works and what doesn't. Someone still needs to stitch it together, polish the edit, layer the music, and ultimately bring the story to life.
At Mediazoo, we understand the future isn't about choosing between human and artificial creation but understanding when each serves us best. What aspects of human creativity do we want to protect, nurture, and combine with these powerful new tools? Perhaps we're asking the wrong question entirely. Instead of "Is this AI or human?" we should ask "Does this serve its purpose? Does it resonate? Does it matter?"
The recognition recession may be inevitable, but what we choose to recognise as valuable, and human - is entirely up to us.