🤔 Why the AI Backlash Isn't About AI

In a blind test, ChatGPT-4's ad copy outscored copy written by senior creatives at one of the world's leading consulting firms. Satisfaction ratings: AI 5.29 vs. Human 4.93. (Zhang & Gosline, MIT Sloan / Berkeley Haas, SSRN 4453958). Pre-registered, funded by Accenture and MIT, not a think-piece. We are now on GPT 5.5.

And yet: Dove, iHeart, Apple TV, and the Authors Guild are spending real money to put "Guaranteed Human" on their content. Why do they pay a premium for a label that, in a blind test, loses?

The popular story is "consumers reject AI." The data says something sharper.

That asymmetry is the whole game, and almost nobody is naming it correctly.

Graphite (May 2026) analyzed 55,400 articles using AI detectors and their findings: roughly 50% of newly published online articles are now majority AI-generated. That figure has plateaued at 50% for five consecutive quarters.

The writing layer is over. Marketers still debating whether AI will "take over content" are debating a 2023 question with 2026 data on the table. Graphite measured production, not consumption, so the share of AI-written content that readers actually see could be lower or higher. But at the creation level, the transition already happened.

The perception wall is going up, not coming down, as consumer sentiment is hardening:

Pew Research (March 2026): 50% of US adults are more concerned than excited about AI in daily life.

Gallup and Walton Family Foundation (April 2026): Gen Z excitement about AI dropped 14 points year-over-year, landing at 22%. Anger rose 9 points to 31%. Hopefulness fell 9 points to 18%.

Fifty-one percent still use AI weekly. Forty-eight percent of working Gen Zers say workplace AI risks outweigh the benefits. This is the generation marketers are spending the next decade targeting.

iHeartMedia (December 2025): 90% of their listeners want media made by humans, including the listeners who themselves use AI. 92% say nothing can replace human connection, up 16 points over 10 years. 96% find "Guaranteed Human" appealing.

And then there's the Merriam-Webster Word of the Year 2025: "AI slop." The culture now has a noun for it. A noun isn't a phase. It's installed vocabulary.

Turn AI Into Your Most Powerful Marketing Teammate

Put AI to Work Across Your Marketing Strategy. In this hands-on course, you will learn how to confidently use tools like ChatGPT, Midjourney, and Descript to plan, create, and scale content across copy, images, and video.

You will leave with practical frameworks and real experience generating marketing assets you can apply immediately. Led by Alec Newcomb, this course is built for marketers, leaders, and business owners who want to improve efficiency, drive growth, and stay ahead with AI.

Many will read this as a transition curve. Familiarity breeds trust. Edelman made that argument in 2025, and it's not wrong: hands-on AI users trust it more than non-users. If the backlash is concentrated among the least exposed, maybe it softens as exposure rises.

But Zhang and Gosline's deeper finding cuts across that logic. The favoritism is asymmetric, and asymmetry doesn't just close.

When subjects were told the same content was human-made, satisfaction rose by b=0.09 (t=2.96, p=0.003) and willingness-to-pay rose by b=0.18 (t=4.14, p<0.0001).

When subjects were told it was AI-generated, the effect was statistically zero. No penalty. Just a missing bonus.

And it's not a quality signal. Tell subjects the content came from "top industry professionals or state-of-the-art AI" without saying which, and the scores flatline. The label has to say "human." Identity, not competence.

Lou et al. (2025, Journal of Interactive Advertising) hit the same wall from the other side: disclose AI use and trust drops, in both the ad and the brand running it. "AI is fine" and "AI is labeled" are different products. One sells at full price. The other gets discounted before the reader finishes the paragraph.

iHeartMedia launched "Guaranteed Human" in December 2025 as a core brand promise across all stations, backed by their own research data.

Dove made the "Real Beauty AI Pledge" in April 2024, becoming the first major beauty brand to commit to never depicting real women through AI generation. They cited projections that 90% of online content would be AI-generated by 2025. They priced ahead of the curve.

The Authors Guild launched "Human Authored" certification in January 2025, expanded it to all US authors in March 2026. Over 3,000 authors have now certified more than 5,000 titles. The seal costs $10 and is trademarked. A certification economy is forming.

Apple TV's Pluribus (Vince Gilligan, 2025) ran a credit line: "This show was made by humans." When Hollywood's marquee names print it on screen, the signal has crossed from defensive positioning to a status signal. Heineken, Cadbury, Brooks Running, and The Tyee are running the same play.

Coca-Cola's AI Christmas ads in 2024 and 2025 were both described as "soulless" across NBC, Marketing Interactive, and the trades. The 2025 version drew worse sentiment than 2024.

Toys R Us ran a Sora-generated ad featuring nostalgic footage of the company's late founder, Charles Lazarus; it was read as graverobbing.

Sports Illustrated published AI-generated content under fabricated author bios; the CEO was fired, and the story is still cited in 2026 disclosure-law debates.

Mango used undisclosed AI-generated fashion models; consumer detection drove weeks of brand discourse.

McDonald's, Klarna, and IBM all publicly walked back customer-facing AI deployments in 2025 after consumer rejection.

The pattern isn't subtle.

The asymmetry isn't just a research finding. It's a business outcome.

ScaledOn produces short-form video for paid social, Reels, and YouTube. Human-led creative, AI-accelerated product.

AI slop and lying are the core of the problem. I’ve seen AI-generated campaigns 4x outperform their counterparts. 

Jin (2025, International Journal of Consumer Studies) complicates the Zhang and Gosline finding in a different direction: algorithm aversion does reappear in unstructured creative tasks once subjects are informed. The disclosure effect may strengthen as AI in marketing becomes more expected, not weaker.

There will be zero gap to close on the AI-penalty side, because there's almost no penalty to begin with. The human bonus looks more durable. It's identity-based, and AI is easy to get angry at. Even as AI output improves (as it does every 21 days as of May), the human label buys something the AI can't: nostalgia for a world that no longer is. Analog is cool again: vinyl, notebooks, old school video games, and landline phones.  

AI scales. Nostalgia doesn't. Human-made is clearly the next vinyl, the question is does the trend last 1 year or 5?

-Alec