Why AI Video Ads Feel Off—and How to Build Trust

An AI video ad can attract attention without earning affection for the brand. Viewers may notice an unfamiliar face or movement, or feel disappointed when an apparent customer experience turns out to be synthetic. Reducing resistance requires designing both the quality of the image and what viewers will understand as fact.

Do people dislike AI advertising itself?

AI ads are not a single category of experience. Kantar reports that ads using generative AI vary in effectiveness; using AI alone does not determine success or failure. [1]

In research published in 2024, NIQ found that the AI ads tested were rated more negatively than non-AI ads. Those findings should not be generalized to every video produced today. [2]

We use this research as a starting point for a production discussion. The categories and workflow below are CREATIP’s recommendations, not universal rules about consumers or guarantees of performance.

What makes an AI video ad uncomfortable?

1. People and objects move against expectations

A still frame may look convincing, while motion exposes problems with gaze, lip movement, contact between hands and objects, and continuity. If fingers slide through a cup or lips fall out of sync, the anomaly may distract from the product. Rather than labeling every defect an uncanny-valley effect, identify the exact moment attention shifts away.

2. A staged scene appears to be a real experience

When a fictional person describes using a product as if they were a real customer, viewers cannot easily judge whose experience they are hearing. Clarify whether a person is a story character or a customer, and whether a location is imagined or real. A small disclosure alone may not undo the impression created by the rest of the video.

3. Product performance and texture become doubtful

An imaginary landscape serves a different purpose from a product demonstration. A change in skin after applying a cosmetic, or a scene showing furniture dimensions and materials, informs purchase decisions. Freely generating these details can leave viewers unsure whether the real product can deliver what they see.

4. The brand’s choices and care are hard to see

Familiar synthetic faces, interchangeable camera moves and grand scenes unrelated to the product offer little reason to remember a brand. Pretesting can also investigate whether the work feels like a cheap replacement for human creativity. The video needs to communicate why this particular expression belongs to the brand, beyond the fact that AI was used.

Which scenes need live action, and where can AI help?

Separate the storyboard into scenes that provide evidence and scenes that invite imagination. The former need verified product and experiential accuracy; the latter can expand the brand’s world. The following is a production decision framework, not a statement of technical limits.

An example: a 15-second cosmetics ad

• —0–4 seconds: paper shapes unfold into an imagined garden to establish the brand’s atmosphere.
• —4–10 seconds: film the actual product and formula to show its packaging, color and spread. Do not synthesize improved skin and present it as evidence of efficacy.
• —10–15 seconds: connect the product to the brand with an end card carrying the same colors and shapes. Keep necessary explanations readable on mobile.

The key is not the percentage of AI footage. It is whether evidence and memorable expression each do their job. Even when an AI background is composited behind filmed product footage, check that the product’s shape, shadow and color remain accurate.

A four-step workflow for more trustworthy AI ads

Reviewing enlarged frames is not enough. Check whether the meaning changes without sound, whether explanations remain readable in a captioned vertical version, and whether shorter edits preserve the distinction between real and synthetic content. Include human review time in the production schedule.

What should you test beyond views?

High views provide limited guidance if people misunderstand the product or cannot recall the brand. Ask these questions just after viewing, before showing the brand name again.

• —Which brand and product is this video advertising?
• —Which scenes did you believe were filmed or based on an actual user’s experience?
• —What felt awkward or difficult to believe?
• —What did you learn about the product?

When comparing revisions, keep the message, duration, audience and delivery conditions as consistent as possible, and specify the variable being changed. For example, compare backgrounds around identical product footage. Research into reactions to AI disclosure is separate from determining disclosure requirements for a live advertisement.

Frequently asked questions about AI video ads

Will disclosing AI use remove resistance?

Disclosure can help explain how content was made, but it cannot guarantee a positive response. The video itself should reduce ambiguity about the person’s role, product effects and real experience. Check current requirements for the country, platform and content type before publication.

Is making the video look non-AI enough?

Visual quality is one review criterion. A natural-looking video can still misrepresent a customer experience or change a product characteristic. Review realism and factual accuracy separately.

Is an entirely AI-generated ad better on a small budget?

Total production costs include revision, editing, audio and review as well as generation. Compare a hybrid estimate that combines essential product filming with AI scenes. Defining what the video must demonstrate makes it easier to allocate the budget.

Trust in AI video starts with designing both how real an image looks and what it asks people to believe is real.

Sources

• [1] Kantar — Rethinking AI-generated advertising: how real people really react
• [2] NIQ — Will AI-generated advertising disrupt the creative industry? (2024-12-12)