Telling consumers that an emotional advertisement was made with artificial intelligence can make the ad seem eerier and less effective. However, that reaction weakens when people can see why AI belongs in the campaign, according to new experimental research.
The study, by University of Mississippi researchers Chang-Won Choi and Robert G. Magee, was published online in the Journal of Interactive Advertising on July 29, 2026. The university drew attention to it in a research release on August 31.
Across two experiments, the researchers examined how viewers responded to AI-generated emotional video advertising. They found that disclosure of AI use could increase perceived eeriness, which reduced the emotional intensity people experienced and weakened their attitudes toward the ad.
The second experiment added an important qualification. When the use of AI fitted the ad’s theme, viewers were more inclined to see the unusual qualities as part of the creative idea. When the fit was weak, they were more likely to suspect that the advertiser was cutting production costs.
Disclosure changes how an emotional message feels
Advertising disclosure usually provides information about how a message was produced or who paid for it. In this case, the label also changed how viewers interpreted what they saw.
The researchers describe the response through the “uncanny valley,” a term first associated with robots and artificial figures that look almost, but not quite, human. Small mismatches in a face, movement or voice can produce discomfort instead of familiarity.
An AI advertisement does not need to contain a robot to create that response. A digitally generated person displaying grief, affection or distress may look close enough to human to invite an emotional reaction while still feeling artificial.
That is particularly relevant to emotional advertising. A product demonstration mainly asks whether the information is useful and credible. A campaign built around compassion or intimacy asks viewers to accept an apparently human feeling as part of the message.
Magee told the University of Mississippi that discomfort can arise when people consciously recognize that a machine does not feel, even though the message appears to suggest that it does.
Why the reason for using AI matters
The study’s broader point is that consumers judge more than the quality of an image. They also form an opinion about the advertiser’s motive.
The authors call this the AI-ad theme match. A strong match exists when the technology helps deliver the central idea, such as creating a genuinely interactive campaign or producing a personalized experience at a scale that conventional production could not easily manage.
In that setting, an imperfect or unusual image may be understood as part of an intentional experiment. If AI is used to imitate work that human actors, illustrators or animators routinely perform, the same imperfection may look like evidence that the company took a shortcut.
“Marketers should start with the campaign idea first and ask, ‘Is AI necessary?’” Choi said in the university’s account of the research.
This does not mean an advertiser should hide its use of AI. The experiment measured the effect of disclosure, but avoiding disclosure may create separate legal, ethical and reputational risks. A more useful response is to make the technology’s purpose clear and to use it where it materially improves the campaign.
Two Coca-Cola campaigns show the difference
The researchers used two Coca-Cola campaigns to explain what they mean by thematic fit.
In 2023, the company launched Create Real Magic, a platform that allowed digital artists to generate work using Coca-Cola’s visual assets and generative AI. The technology was part of the activity offered to participants, rather than simply a production method operating out of sight.
Coca-Cola also used AI to rework its familiar “Holidays Are Coming” advertising in 2025. The company described the campaign as AI-powered storytelling, but it also attracted online criticism from viewers who questioned the production method and the appearance of some images.
The contrast is useful, but it needs to be handled carefully. The study did not conduct a controlled test comparing the commercial performance of those two campaigns. Online mockery is not the same as a measured fall in sales, and neither example proves that AI alone determined the result.
Instead, the campaigns illustrate the question at the center of the experiments. In Create Real Magic, the audience could readily understand what AI added. In a conventional animated commercial, the reason may be less obvious.
The study does not cover every AI advertisement
The results concern emotional video ads shown in experimental settings. They should not be generalized automatically to product images, search advertising, personalized offers, business-to-business campaigns or AI used behind the scenes for editing and media planning.
The outcome measured was viewers’ response to the advertisement. It was not a direct measure of sales, customer retention or long-term brand value.
Other research nevertheless points in a similar direction. A separate 2026 study in the Journal of Retailing and Consumer Services reported that AI disclosure could reduce attitudes toward an advertisement through perceived eeriness, although a stronger brand could soften the effect.
Consumer expectations may also change. Computer-generated imagery and animation once drew attention simply because they were new, but both are now routine. Magee said AI could follow a similar path as people become more familiar with it.
At the moment, the safer commercial test is whether AI gives the audience something that the campaign could not readily provide without it. If the answer is unclear to the advertiser, it may be even less clear to the customer.
