I didn’t expect the reaction we got after the Emily in Paris piece. Not because the criticism was unfair — it wasn’t — but because it exposed a much bigger tension that everyone in the media is currently navigating and rarely talking about honestly.

We were criticised for using AI-generated illustrations (on Instagram) while analysing a show built by real people — costume designers, set decorators, location scouts, entire production teams whose work exists precisely because it is human. The accusation was simple: hypocrisy.

It is uncomfortable. And it is also the right place to start a real conversation about AI fatigue — not opinions, but facts.

Let’s begin with something basic that often gets lost in the noise: AI-generated content doesn't eliminate human labour. It displaces visible labour, not cognitive labour. When our art director works with AI, she doesn’t “generate art.” She builds a concept, tests dozens of prompts, rejects outputs, manually corrects distortions, adjusts tone and references, and aligns visuals with editorial meaning. According to Adobe’s 2023 Future of Creativity report, over 74% of professional creatives using generative AI report increased time spent on ideation and refinement, not less. AI shifts effort upstream — it doesn’t remove it.

But here is the part that matters more than process: audiences don’t care how hard the tool was to use. They care how it feels. And right now, audiences feel saturated.

A 2024 Nielsen Norman Group eye-tracking study on social media consumption showed that users spend 37% less time engaging with “synthetic-looking” imagery compared to content perceived as authentic or user-generated — even when the informational value is identical. The brain flags AI-heavy visuals the same way it flags ads: fast recognition, fast dismissal.

This is especially visible on Instagram.

Meta’s own internal research, partially disclosed during the 2024 creator monetisation briefings, showed a measurable decline in completion rates for AI-heavy Reels, particularly in wellness, fitness, and lifestyle categories. At the same time, Instagram’s algorithm began explicitly boosting “low-production, original camera content” — a shift confirmed by Adam Mosseri himself in multiple public Q&As in late 2023 and early 2024.

The wellness niche is the clearest example of where AI fatigue becomes actively harmful. Nutrition advice delivered by bodies that don’t exist. Training routines demonstrated by anatomically improbable physiques. Mental health guidance illustrated by faces that have never experienced stress. According to a 2024 Stanford Digital Health Lab review, synthetic influencers in health-related content reduce perceived credibility by up to 41%, even when audiences can’t explicitly identify the content as AI-generated.

People don’t trust perfection anymore. They associate it with manipulation.

That is why one of the fastest-growing content trends on Instagram, TikTok, and YouTube Shorts in 2024 and 2025 is what platform analysts call “deliberate de-polishing”: handheld footage, imperfect lighting, no makeup, no filters, no post-production. A Harvard Business Review analysis of creator trust metrics showed that audiences rate “unfiltered human content” as 2.3x more trustworthy than high-production visuals in lifestyle and fitness categories.

I see this firsthand in cycling. Endurance sports aren't cinematic. They are repetitive, uncomfortable, sweaty, and often ugly. And that is precisely why real footage works. I don’t care if my hair looks good. I don’t care if I look tired, red, or dirty. That isn't branding — that is physiology. And people recognise it instantly.

Now, contrast this with how big brands behave.

Apple doesn't use AI to replace creatives. It uses AI to support them. Its Shot on iPhone campaigns are built entirely around real photographers and filmmakers, with Apple acting as a patron rather than a generator. Hermès still commissions illustrators for scarves that take months to develop. Jacquemus builds physical sets, real locations, absurdly expensive installations — because it understands that luxury is inseparable from human effort.

The luxury sector has one of the highest creative labour investments per campaign, precisely because it understands that cultural capital can't be automated. A 2023 Bain & Company report on luxury brand equity showed that brands investing heavily in human-led creative production retain customer loyalty 18–22% longer than those leaning into automation-heavy visual strategies.

But media companies don’t operate on luxury margins.

This is the part critics often ignore. Independent media doesn’t have the budget to commission illustrators, photographers, and designers for every article. Creative labour is expensive because it should be. Artists invest years into their craft, and fair compensation reflects that. But not every story can carry that cost.

This is where AI becomes acceptable — as a tool, not a disguise.

The ethical line isn’t about using AI at all. It is about using it to simulate human experience — especially in areas where trust matters: health, fitness, mental wellbeing, lived cultural commentary. When AI pretends to be human, it erodes credibility not just for one post, but for the entire ecosystem.

And the data supports this shift. Across platforms, across industries, across geographies, the same signal repeats: audiences don’t want less technology — they want less deception. They want to know when something is illustrative, when something is symbolic, and when something is lived.

The next phase of content isn’t anti-AI. It is anti-fake authority. Being human — tired, imperfect, inconsistent — is no longer a weakness. It is the only thing that cuts through.

And maybe that is why this conversation is so tense. Because it forces us, especially in the media, to stop hiding behind aesthetics and ask a harder question: are we producing images, or are we communicating experience?

AI can help us create faster. It can't help us be honest. And right now, honesty is the only thing people still stop scrolling for.