Analysis by E.H. Bradford

AI Industry Reporter & Reality Correspondent

AI Video's Reality Check: How Big Brands Are Getting Burned—and What Founders Can Learn

AI video has become marketing's shiny new toy—and a few short years in, it's already exposing who actually understands storytelling and who's just feeding prompts into a machine. The same tools that help scrappy teams punch above their weight are also turning billion‑dollar brands into cautionary memes.

Underneath the headlines about "AI slop" and ad backfires is a simple pattern: when AI supports a clear human idea, it quietly prints results; when it becomes the star of the show, the audience feels used.

When AI Video Works: Assist, Don't Replace

The success stories don't start with "we wanted to try AI." They start with "we already knew what we wanted to say." In those cases, AI is less replacement and more power tool—loud, fast, and very effective when you point it at the right job.

One global brand ran controlled tests pitting AI‑made ads against UGC and designer‑built spots. The AI variants were far cheaper and faster to produce and, on simple product‑driven messages—"here's what this thing is, here's what it does"—some of those machine‑assisted ads matched or beat the human‑crafted ones on engagement and recall. The key detail: humans still owned the brief, the promise, and the final decision; AI handled variations, formats, and extra cuts, not the soul of the idea.

Agencies working this way talk about AI like an assistant director: it generates B‑roll, fills in stylized backgrounds, and creates transition shots they could never justify in a low‑budget timeline. Human actors and clear arcs stay at the center; the models simply widen what's possible visually on startup‑sized budgets.

When you zoom in on these wins, they all share one quiet rule: the story is written by people, then multiplied by machines.

When AI Becomes the Ad: McDonald's, Coca‑Cola, and the "Slop" Era

The trouble starts when brands flip that formula—when the story bends around what the model can do, instead of the model bending around what the story needs. That's how we got a new kind of category: AI ads that are famous purely because they went wrong.

McDonald's Netherlands: Christmas Without the Comfort

McDonald's Netherlands walked straight into that wall with a Christmas spot built almost entirely from generative clips. The ad stitched together a rapid montage of icy bike crashes, traffic jams, burnt dinners, chaotic family scenes and other catastrophes under a twisted version of "It's the Most Wonderful Time of the Year," reframed as "the most terrible time of the year." The closing move: if the holidays are unbearable, you can always escape to McDonald's and hide there until January.

Viewers didn't just shrug—they recoiled. Social feeds filled with words like "creepy," "soulless," "AI slop," and the now‑headline‑famous line: "it ruined my Christmas spirit." People called out uncanny faces, stiff movement, background glitches and a cold, mechanical feel that clashed with what a December ad from a family‑positioned brand is supposed to deliver. Within days, McDonald's pulled the ad, disabled comments and labeled the experience "an important learning" in how to use AI responsibly.

Analysts who went frame by frame pointed out something more subtle: the narrative had effectively been shaped around AI's current strengths and weaknesses—lots of one‑second shots, few close‑ups, minimal continuity—rather than around a coherent emotional journey. Instead of "we understand your December chaos and offer a small comfort," viewers heard "we pointed a generative model at your holidays and hit publish."

Coca‑Cola: A Classic Reboot That Lost Its Heart

Coca‑Cola's holiday work shows a softer version of the same problem. For decades, its "Holidays Are Coming" truck ads traded on pure nostalgia: real trucks, snow‑dusted towns, that instantly recognisable jingle. In recent years the company leaned into generative AI to re‑imagine those scenes—AI‑rendered trucks, villages and characters that closely mimic the original.

On paper, it was smart: same structure, modern technique. In practice, viewers spotted trucks that morphed between shots, gained or lost wheels, and characters whose faces and movements felt subtly plastic. Critics described the result as "a cover band of its own history"—technically polished, but emotionally thinner. Yet effectiveness tests still showed strong recall, proving a blunt point: you can hit your key metrics and still erode the intangible trust that made those ads iconic in the first place.

Fashion's AI Models and the Human Vacuum

Move away from Christmas and the same pattern shows up in fashion. Brands including Mango, Levi's, H&M, Skechers and Guess have experimented with AI‑generated models in campaigns and lookbooks, promising diversity and efficiency while cutting production costs.

Audience reaction hasn't been kind. Even when the images look superficially slick, people call out uncanny poses, too‑perfect bodies and the feeling that real photographers and models are being quietly deleted from the frame. Industry numbers tell the story: brand deals with AI "influencers" fell sharply in 2025 as marketers pulled back, while labels like Heineken, Aerie and Polaroid leaned publicly into "no AI" positioning as a way to signal human‑first values.

Across these cases, the pattern is clear. The backlash isn't against any use of AI; it's against campaigns where the technology feels like the real protagonist and the human point of view is missing in action.

What Research Says About AI‑Driven Storytelling

Step back from individual brands and you start to see the same narrative issues mapped out in research. UX analysts, ad effectiveness studies and industry post‑mortems are all circling the same conclusion: most failing AI ads are software demos disguised as stories.

1. "AI First" Concepts Break the Story Spine

One detailed analysis of AI‑generated holiday ads notes that many campaigns were built around the tool, not the viewer: teams wanted to show off one‑second generative clips, extreme camera moves and surreal transitions, then retrofitted a loose "holiday chaos" theme on top. The result was a stitched‑together reel of what the model could render, rather than a coherent emotional arc a human would naturally tell.

By contrast, high‑performing emotional ads—whether or not AI is involved in production—start with a simple human premise (loneliness, hope, relief, reconnection) and only then decide which scenes and tools serve that feeling. The structure is old‑school: a situation, a shift and a resolution where the brand has a clear, believable role.

2. Fragmented Visuals Need Stronger Narrative Anchors

Generative models love quick, varied shots: they hide continuity flaws and make it easier to avoid those cursed close‑ups where eyes and hands go wrong. That's why so many AI ads lean on fast montages and distant framing. But the more fragmented the imagery, the more important it becomes to give viewers something stable to hold onto—a recurring character, a clear voiceover, a simple before/after arc.

Coca‑Cola's better‑received AI work retains a recognizable skeleton: the trucks, the jingle, the journey through town, the final shared moment. Even when details wobble, the audience can predict where they are in the story. The worst offenders swap that arc for a series of "cool shots," and people respond exactly as you'd expect: intrigued for a second, unmoored after ten.

3. High‑Arousal Negative Emotion Travels—for the Wrong Reason

Research on virality has a consistent finding: high‑arousal emotions—like awe, anger and anxiety—drive sharing more than low‑arousal ones. AI ads that lean on chaos, disgust or fear often rack up views for exactly this reason. But zoom into the comments on campaigns like McDonald's Netherlands or the more jittery AI Christmas work from other brands and you see the catch: people are sharing to ridicule or criticize, not to celebrate.

The narrative structures that actually boost long‑term brand sentiment pair tension with relief: messy dinner then laughter and connection; winter chaos then a small, earned moment of calm. AI‑heavy ads that stay stuck in the "everything is terrible" phase without making that turn feel less like empathy and more like contempt.

4. Visible Human Authorship Still Matters

Interviews and surveys around these campaigns show a consistent undercurrent: audiences want to feel that a real team with a specific point of view made editorial choices. That doesn't mean every frame must be shot "analog," but it does mean the work should feel authored, not autogenerated.

People pick up on cues: real faces and voices at key beats, little imperfections that don't look like training‑data noise, context about how and why AI was used. In contrast, work that feels like something any brand could have generated with the same prompt often gets written off as lazy, regardless of technical sophistication.

Narrative Best Practices for AI Video (That Don't Require a Big Team)

The good news for solo founders and small teams is that the guardrails here are narrative habits, not expensive tech. You can borrow the same discipline big brands are relearning the hard way.

1. Start With the Human Logline

Before you open an AI tool, write the one‑sentence human version of your video: "A burned‑out creator finds a 30‑second breather with our app," "A confused shopper becomes confident in three steps using this template," "A local client goes from invisible to booked out." If that line sounds petty, confusing or cruel when you say it out loud, no model can save it.

This does two things. It forces you to pick a protagonist (not "everyone"), and it forces you to choose a feeling you want to leave them with. That's the spine. Everything else is decoration.

2. Map a Simple Emotional Path

On one page, sketch three beats: starting state, friction, resolution. "She's overwhelmed, she discovers a simpler way, she's in control again." "They can't get clients, they change their approach, the first booking comes in." It's basic story craft, but it's also the antidote to AI‑driven chaos reels.

Once that arc exists, you can decide where AI belongs. Maybe it generates a stylized backdrop for the "friction" section, or quick cutaways that show your product in clever contexts. What it shouldn't decide is whether your story ends in despair or relief.

3. Use AI to Fill Space, Not Define Turning Points

Use generative tools for B‑roll, scene transitions, text treatments, and alternate edits of moments you've already planned. Keep the key beats—problem reveal, moment of change, resolution—anchored in either real footage or very carefully reviewed AI output.

If you notice that every big moment in your video only exists because "the model did something crazy there," that's a warning sign. You're building around novelty, not narrative. That's exactly how you end up in McDonald's territory: the AI is the punchline, and your brand is the joke.

4. Respect the "AI Tells"

Hands, eyes, logos, text—these are the areas where models still stumble, and viewers have learned to spot those artifacts instantly. Treat obvious glitches the way you'd treat typos on your homepage: fix them when you can, and if you choose to leave one in for stylistic reasons, do it consciously.

Not every video has to be flawless, but when a detail pulls the viewer out of the story and into "wow, look how weird that hand is," your message has already left the building.

5. Run a "First‑Time Viewer" Check

Before you post, watch your own video like a stranger and answer three questions without pausing: What's happening? How am I supposed to feel? Why is this brand part of this moment? If you can't answer one of those in plain language, you're asking your viewer to do work they won't do.

Then look at your most AI‑heavy shots and ask a fourth: if someone screenshotted this frame and shared it with the caption "look at this AI ad," would you be proud or embarrassed? If the answer is anything other than "fine with it," go back and trim.

6. Protect Your Voice Like an Asset

Have a simple voice guide written down: a few phrases you use, a few you avoid, your usual emotional range (calm, hype, dry, playful), and topics you handle with extra care (money, mental health, holidays, identity). When AI suggests scripts or visual ideas, run them through that filter instead of assuming the defaults are safe.

This doesn't just protect you from McDonald's‑style tone‑deaf moments; it keeps your work from blending into the generic "marketing‑speak" that models have been trained to mimic.

Where This Leaves Founders, Creators, and Small Teams

Strip away the budgets and the PR disasters, and the underlying rule is surprisingly egalitarian: AI video amplifies whatever discipline you already have. If your storytelling is tight and your empathy is switched on, it lets you test more ideas, in more formats, on more platforms than you ever could alone. If your ideas are half‑baked or your connection to your audience is thin, it just makes those problems louder.

For solopreneurs, freelancers and small businesses, that cuts both ways. You can now ship a product explainer in an afternoon, build a mini‑campaign around a new template or offer by the weekend and keep up a steady stream of short‑form content without hiring a crew. You can also, with a few bad prompts and no review, land yourself in a thread about "AI slop" before your next launch even starts.

The companies getting burned right now aren't dumb; they're rushed, dazzled and a little too eager to let the tool steer. If you can resist that impulse—if you keep the human logline, the emotional arc and the final call firmly in your hands—you can use exactly the same technology to do what small players have always done best: tell sharper, more specific, more honest stories than the big guys can manage.

AI doesn't change the stakes. It just accelerates the reveal. When your next video hits the feed, it will tell people who you are faster than ever—whether you meant it to or not.