82% विज्ञापन अधिकारियों को लगता है कि युवाओं को AI विज्ञापन पसंद हैं। वास्तव में 45% को पसंद हैं।

In a study published in January, IAB asked US ad executives and Gen Z and Millennial consumers about the same thing. 82% of the executives believe those consumers feel positive about AI-generated ads. 45% of the consumers say they do. The gap is 37 points, and IAB says it was 32 in 2024. Add two holiday ads that drew public backlash and an AI-generated "actress" who slipped into Cantonese in the middle of a September interview, and the question deserves a proper answer: what exactly are people objecting to?

We make AI ads for a living, so we have a stake in the answer. Where what follows is our opinion rather than a reported finding, it says so. Every outside fact is listed in the sources at the end.

The number, and how it was measured

IAB ran the study with Sonata Insights between October 2025 and January 2026. It surveyed 505 US consumers, Gen Z aged 16 to 27 and Millennials aged 28 to 43, and 104 US ad industry executives.

• 82% of executives think Gen Z and Millennial consumers feel very or somewhat positive about AI-generated ads.
• 45% of those consumers actually do.
• The gap: 37 points in 2026, against 32 in IAB's 2024 study.

Read it with its limits. It is one survey, US only, with a modest sample, and it measures what people say they feel, not what they watch or buy. It is still the cleanest side-by-side we know of, because both groups answered.

Who objects most

Gen Z, by a distance. 39% of Gen Z respondents feel negative about AI-generated ads, nearly double the 20% of Millennials. In IAB's earlier study those figures were 21% and 15%, so the younger group has moved further and faster. Asked to describe brands that use AI for ads, Gen Z chose these words far more often:

• Inauthentic: 30% of Gen Z, 13% of Millennials.
• Disconnected: 26% and 8%.
• Unethical: 24% and 8%.

If your customer is under 28, you are talking to the most skeptical audience in the survey.

The words each side uses

IAB also asked both groups to describe brands that use AI to generate advertising. They could have been describing different brands.

• Innovative: 49% of executives, 23% of consumers.
• Forward-thinking: 46% and 22%.
• Unique: 44% and 26%.
• Manipulative: 10% and 20%.
• Unethical: 7% and 16%.

Executives choose words about the brand's ambition. Consumers are about twice as likely to choose words of distrust. That is the gap in one line.

Three moments that set the mood

In November 2024 Coca-Cola's AI remake of its 1995 "Holidays Are Coming" spot was called "soulless" online, NBC News reported. In December 2025 McDonald's in the Netherlands removed a 45-second AI Christmas film within days of posting it, and the chief executive of the production company wrote that the job had taken ten people, full-time, for five weeks. In September 2026 Gizmodo reported that Tilly Norwood, an AI-generated character made by the London studio Particle6, lagged, gave flat answers and switched into Cantonese during a YouTube interview.

Neither holiday film is a story about software failing to render a frame, and the McDonald's film had real craft hours behind it. In our reading, what people reacted to was a feeling: that something meant to be human had been handed over without enough care. We look at the two holiday films closely in filmito.io/en/blog/ai-holiday-ad-2026-lessons.

The counter move: pledges about people

Some brands now market the absence of AI. Dove committed in April 2024 to "never using AI in place of real people in its advertising". Aerie pledged in October 2025 not to use AI-generated people or bodies, and its chief marketing officer, Stacey McCormick, told Marketing Brew that "anything that replicates a body or a person is absolutely not" acceptable, while AI for planning is "more acceptable".

Notice what both pledges cover. Neither brand promised to stop using software. They drew the line at the human being on screen. For anyone briefing an AI ad, that line is the most useful thing in this post.

What does not seem to hurt: telling people

In the same IAB study, 73% of the consumers surveyed said that knowing an ad was made with AI would raise or not change their likelihood to buy, and consistent disclosure is one of IAB's three recommendations. YouTube, which announced in May that it would label videos automatically when its systems detect significant photorealistic AI that a creator did not disclose, says that "a disclosure label alone does not change how a video is recommended or whether it's eligible to earn money".

The reaction online is not the whole picture either. System1, a firm that tests ads on viewer panels, gave Coca-Cola's 2025 AI holiday film 5.9 stars, the maximum its model awards, and concluded that "using AI doesn't hurt ordinary viewers' enjoyment of the ad at all". It added a caveat worth keeping: a few negative comments suggest the brand was wise not to mention AI in the ad itself. In our view, a plain label and making AI the message are different things.

Our reading: people object to carelessness, not to the tool

This section is opinion, from a studio that directs AI films for clients and runs four engines in production, chosen shot by shot, as we describe in filmito.io/en/blog/seedance-vs-kling-vs-veo-vs-grok-for-ads. In our experience the engine is never the reason an audience trusts a film.

What we see in the AI ads that get mocked is a list of small neglects. A product that is almost right. Faces with nothing behind the eyes. A story nobody wrote. Music laid under the pictures at the end. Each one says "nobody checked", and an ad nobody checked is easy to read as manipulative. IAB's second recommendation says the same thing in industry language: use AI "to enhance creative quality, rather than focusing on accelerating production of advertising assets alone".

That is why a person signs off every frame we deliver. One client, alliyah30 in the United States, put the standard in a Fiverr review better than we could: someone "who will treat your work as their own". The gap closes when the people making the ad care about it as if it were theirs.

Filmito is an independent studio and a vetted Fiverr Pro seller. We are not affiliated with, endorsed by, or sponsored by Fiverr. Fiverr is a trademark of its owner.

Six production choices that close the gap

• Real product references. Photographs of the actual product, front, side and detail. The pack on screen has to match the pack on the shelf, and prices and claims are set as type in the edit, because text inside generated footage garbles.
• A written story before anything is generated. One character, one turn, one reason to remember the brand. Our Sanabelle film follows a cat through a run of doors, gluten-free or not, 2 kg or 10 kg, sugar-free or not, and ends with the cat between the two packs. Watch it at filmito.io/en/case-studies?play=native-boch-sanabelle-v1.
• Designed sound. On our projects the client approves the voice and the music before any footage is made. Sound chosen last is the fastest route to generic.
• A face only when it earns its place. Generated people are where viewers look hardest and where models drift first. If the idea works with hands, objects, animals or places, let them carry it.
• A reviewer who was not in the room. Someone who has not read the brief watches the cut on a phone. If they notice the AI before they notice the product, the film is not finished.
• A label when a synthetic person appears. Since June 9, 2026, New York has required a conspicuous disclosure on ads in the state that feature an AI-made performer, one who is not any real, identifiable person, when whoever creates or produces the ad knows it is there. Elsewhere IAB recommends disclosure, and the 73% finding suggests it costs little.

This is general information, not legal advice. Disclosure rules differ by market and platform, so check yours with counsel before launch.

Two tests for your own ad

• Watch it muted. If the product and the story still land without the music, the film is doing the work, not the soundtrack or the novelty.
• Picture it with an AI label in the corner. If that makes you uncomfortable, ask why. Usually something in the film is pretending: a customer who never used the product, a generated person presented as a real one, a place passed off as somewhere it is not. Fix that and the label stops being a worry.

What this survey cannot tell you

It does not measure sales, recall or watch time. It does not separate a fully generated film from a real product shot with a generated background, and those may be received very differently. It is a snapshot of one market. Treat the 37 points as a warning about assumptions inside the industry, not a forecast of results. For the longer argument about where AI helps the craft and where it hurts, see filmito.io/en/blog/will-ai-kill-real-filmmaking and filmito.io/en/blog/we-direct-ai-we-do-not-generate.

If you want an AI film made with these choices built in, send the product and the idea through filmito.io/en/see-your-ad. We reply within one business day with the idea, the price and the date. Our approval data, including the projects that took longest, is public at filmito.io/en/ai-ads-report.