At the 73rd Cannes Lions, which opened June 22, less than two weeks ago, the advertising industry had a rare moment of collective honesty. Cannes CEO Simon Cook described what three years of generative AI adoption had produced in three words: "mediocrity at scale." He wasn't editorializing. He was reading the data. AI-identified content now gets 20 to 35 percent lower engagement than comparable human content. Consumer enthusiasm for AI-generated material has collapsed from 60 percent in 2023 to 26 percent today. Cannes entries fell 25.5 percent this year, the first meaningful drop in years, and the festival responded by requiring every submission to be personally endorsed by the submitting agency's CEO and CMO. The production revolution worked. The audience didn't follow.

The paradox nobody in the room wants to name

The gap is that spending and results now move in opposite directions. In 2026, 79 percent of marketers increased their AI content investment. At the same time, 54 percent of consumers say they are annoyed by how much they hear about AI, and 65 percent want brands to stop mentioning "meaningless AI marketing" altogether. McKinsey tracked 90 percent of CMOs actively experimenting with AI in their marketing, but found fewer than 10 percent have scaled it or captured real value. The money flows one direction; the returns go the other. This is not a temporary lag while audiences adjust. It's a structural mismatch between what marketers believe and what human attention actually rewards, and it widens every quarter.

Your brain rejects synthetic content before you know it's synthetic

Your nervous system flags fake before your conscious mind ever names it. This is the layer most media plans skip entirely. For two hundred thousand years, human survival depended on reading whether another person was being genuine. Is this stranger actually friendly, or performing friendliness? The brain structures that handle this, the amygdala and the anterior insula, did not disappear when we got smartphones. They run continuously in the background, pattern-matching against subtle signals: the slight inconsistency in a story that flags rehearsal, the smoothness that signals construction rather than lived experience, the frictionlessness that tells your nervous system nothing was at stake when this was made. When content triggers that read, researchers call it the uncanny valley of intent. We don't consciously register the signal. We just feel a vague resistance to trust the brand. Scroll immunity is not apathy. It's an ancient detection system running at 60 frames per second on a modern feed.

When anyone can produce a competent asset in seconds, competence stops being valuable. Taste becomes the only premium left. (Cannes Lions 2026 jury observation)

Authenticity is not a brand value. It's an economic variable.

Treat authenticity as a line item, because it moves your CPM. Run the math. If AI content gets 35 percent lower engagement, your effective CPM is roughly 54 percent higher than the headline rate, because you're paying to reach people who are not acting. Micro-influencers are outperforming mega-creators by 20 percent on conversion rates precisely because they carry the friction of a real personality: an opinion that rubs someone slightly wrong before it builds trust, a camera angle that's slightly off, a joke that doesn't quite land. These are not production errors. They are trust signals. The brands shifting spend from volume to signal are seeing measurable lifts in the metrics that drive actual revenue: hold rate, saves, and direct search spikes after exposure. Publicis CEO Arthur Sadoun made the same argument at Cannes this year by releasing a satirical film calling out AI overpromising in agency pitches. The punchline landed because every CMO in the room recognized the pitch.

What this means for your next production cycle

Keep AI, but take back the one job it can't do. The answer is not to abandon AI. It's to stop outsourcing the human editorial layer that decides what gets made, which angle it takes, and whether the result sounds like a person with a point of view or a prompt with a fill-in-the-blank tone. Three moves that cost nothing to start: audit your last quarter of content and ask honestly which pieces feel human and which feel machine-smooth. Run a test, unpolished against polished, on your best-performing format; the results will challenge every assumption you have about production quality. And find the one voice in your organization or creator network that carries genuine authority in your category. Point your AI spend at amplifying that voice, not at generating volume around it. Volume is what broke Cannes. Voice is what wins business.