AI Images or Real Photos?

1. Brand hypocrisy: the claim of social values doesn’t align with the synthetic imagery

The strongest evidence in this area comes from a study conducted by Quan Xie, Sidharth Muralidharan, and Joe Phua of the Temerlin Institute of Advertising at Southern Methodist University, which Journal of Advertising'The study was published in [...]. The research team began examining this issue following Levi's announcement that it would use artificial intelligence models to increase diversity and the wave of backlash it faced.

In a three-phase series of experiments, the use of AI-generated human models in body positivity and plus-size apparel campaigns led to a significant brand hypocrisy It created a certain perception. Result: Attitudes toward the brand, purchase intent, and the likelihood of recommending it all decreased significantly.

Mechanism Social Presence Theory It works through this mechanism. As researcher Xie puts it, consumers do not perceive AI models as human, warm, or relatable. When a brand that champions human-centered and empathy-driven values uses synthetic imagery instead of real people, it undermines the credibility of its message.

A parallel finding, by Brüns and Meißner's Journal of Retailing and Consumer Services'It comes from a 2024 study published in: the use of artificial intelligence in social media content directly undermines the perceived originality of the brand.

2. Signal theory: Visuals convey not only the product but also the seller’s effort

Research based on Signaling Theory shows that marketing visuals do not merely show the product to the consumer, but also the effort and care shown by the seller about claims to provide information. A series of three experimental studies conducted on food and product images shows that images labeled as "AI-generated" influence consumers' perceived merchant effort (perceived merchant effort) shows that it weakens the inference.

When consumers perceive that a seller has taken the easy way out in the production of visual content, they also lower their expectations regarding the product’s quality; this directly reduces their intention to purchase. The impact is significantly more devastating for products commanding a high price premium. In other words, the cost of an AI-generated image is higher for a premium brand than for a budget brand.

3. The level of interest in a product plays a regulatory role

Experimental design studies in which the source of the advertisement (artificial intelligence or human) and the status of transparent labeling are tested, level of product involvement This shows that it is a decisive factor. For low-interest products (such as snacks), an AI-generated image is met with less resistance; whereas for high-interest, high-financial-value, or technical risk (laptops, luxury goods), a clear “AI” label sharply reduces trust and purchase intent.

This finding forms the academic basis for the categorization matrix: As risk increases, the return on actual production increases. The same budget yields a low marginal benefit when spent in a low-interest category, but a high marginal benefit when spent in a high-interest category.

4. The rationale is more decisive than the technique itself

This is the most useful finding from a business perspective. It reveals what motivates brands to use artificial intelligence to reach consumers how he explained it In an experimental study examining its effect on trust (n=209), rationale “privacy protection” When presented with these images, consumers showed similar levels of trust as they did with computer-generated images. The same technology “cost-effectiveness” When explained on these grounds, statistically significant declines in confidence and purchase intent were recorded.

5. Field test: AI attracts attention, real photos sell

The evidence that most directly supports the thesis of this article comes from a field experiment conducted on a C2C (consumer-to-consumer) marketplace. On the Vinted platform, AI-enhanced product photos were compared with the original photos. The result is twofold:

• AI won the attention phase. As the visual appeal increased, click-through and like rates went up.
• The original photo was retained during processing. When it came to final conversions and purchases, unretouched, real photos proved to be the deciding factor.

The conclusion is clear: In C2C marketplaces—and increasingly across all e-commerce—the primary function of imagery is not aesthetic perfection, but, It serves as evidence documenting the product's actual condition. Artificial intelligence increases transaction risk because it erodes this evidence-based function. This aligns perfectly with the cart abandonment and return data from the previous section: gains at the top of the funnel, losses at the bottom.

6. Tolerance varies depending on the type of image

A study conducted by the Münster School of Business (FH Münster) and CECIRE in partnership with the fashion retailer Ernsting's family found that consumers in decorative and background images that it shows a certain degree of tolerance toward artificial intelligence; however, in photos of personalized products and models He confirms that he takes an extremely critical stance toward the use of AI. This finding forms the basis for the “decision table by image type” presented later in this article.

7. Other Supporting Findings

• “Even the word ”AI" can lower purchase intent A study by researchers Cicek, Gursoy, and Lu at Washington State University shows that the appearance of the term “Artificial Intelligence” in product and service descriptions has a negative effect on purchase intent. The mechanism operates through emotional trust; the effect is more pronounced in high-risk categories such as expensive electronics, financial services, and medical devices.
• Perceptions change once the source is revealed. In Zhang and Hur’s 2025 study, when the source of an image is not disclosed, there is little meaningful difference between AI-generated and human-generated images. However, when it is disclosed that an image was generated by AI, trust and purchase intent drop significantly.
• The risk is higher in hedonistic categories. Belanche et al. International Journal of Information Management'A study published in [...], shows that the use of AI in visual contexts is perceived more negatively in hedonic and high-engagement decision-making processes.
• A moral revulsion response to emotional content. Journal of Business Research'A study published in [journal] shows that when consumers believe emotional marketing messages were written by AI rather than by humans, they find them less authentic, feel moral aversion, and their purchase intent weakens — even though the content is exactly the same.
• An ethical stance helps soften the reaction. Four experimental studies conducted at Swinburne University of Technology under the leadership of Prof. Sean Sands show that negative reactions largely disappear when a brand demonstrates a credible stance on social good and ethical responsibility.
• “The assumption that ”they will get used to it in time" is weakening Data tracked by Conjointly between 2023 and 2025 shows that acceptance of AI-generated content does not automatically increase as technology advances, but has actually declined in some areas. Fractl’s 2026 data reinforces this warning: the trend is not toward acceptance but toward a sharp shift.

In e-commerce, the visual replaces the physical product at the point where the consumer cannot touch the product. Fabric texture, metal shine, glass permeability, leather surface, cosmetic pigment, jewelry size, packaging quality, Details such as how the product looks on the hand or body are read on the visual.

For this reason, if an AI-generated product image depicts the product as more flawless, brighter, smoother, larger, of higher quality, or in a different color than it actually is, it creates a commercial risk. Even if the image is aesthetically pleasing, it raises the following question in the consumer’s mind: “Is this really how the product will arrive?”

In the luxury, cosmetics, and fashion categories, however, there’s an additional layer. The problem highlighted by the study “When AI Doesn’t Sell Prada” is not merely a matter of reality: in luxury consumption, value is often tied to human labor, craftsmanship, attention to detail, materials, and the story behind production. When AI-generated imagery undermines this perception of labor, consumers begin to question the brand’s approach to value creation.

Data from 2026 brings this finding to the commercial level. In a survey conducted by Clutch in June 2026 with 408 consumers, the participants %36'sı markanın arkasında gerçek insanlar görmenin sadakatlerinin en güçlü sürükleyicisi olduğunu says — it’s more powerful than price, product quality, and convenience. In the same study, identifies real customer stories as the most memorable campaign format.

Same image, different label = different level of trust

A study by the NIM (Nuremberg Institute for Market Decisions) shows that labeling the same image as a “photo” or “AI-generated image” changes consumer perception. An image labeled as “AI-generated” is perceived as less emotional, less credible, and less memorable. This finding reveals that transparency does not always build trust; in some cases, it brings consumer skepticism to the surface.

However, the third experiment in the SMU study rounds out this picture: While a simple “Generated by AI” label may erode trust, a well-crafted transparency statement explaining how the process works enhances brand perception. For example, stating that the AI model was trained on a dataset of photographs of real people for whom consent was obtained and who were licensed helps alleviate consumer skepticism.

Rule: AI should never be presented as a substitute for a value commitment

This is the one-sentence rule that e-commerce brands can apply directly. Claims of diversity, body positivity, sustainability, local sourcing, handmade products, or social responsibility —when supported by stock images—in the consumer’s mind brand hypocrisy is triggered. Technology must be clearly positioned as a tool that supports operational speed and creative processes.

In categories such as beauty, baby care, food, luxury, apparel and analog photography, the message “we don't use AI” is becoming increasingly visible. This is because in these categories, the purchase decision is not only based on price and function. Body reality, tactility, human representation, the physical feel of the product and the brand's approach to labor are all part of the decision.

Dove’s commitment not to create or distort images of women using AI, Aerie’s policy of not using AI-generated bodies or people, and Coterie’s decision not to use AI-generated social media images , and Polaroid’s campaigns highlighting the physical experience all point to the same commercial reality: In the age of AI, physical reality is becoming a brand asset again.

In 2026, this approach evolved from a campaign-level initiative to a brand commitment. Aerie expanded its '100% Aerie Real“ commitment through a campaign featuring Pamela Anderson; the ad’s concept tackles the issue head-on—Anderson has an AI tool generate models, tries to tweak the results to make them ”more natural, less lifeless,“ eventually gives up and moves to a real set, where she turns to the camera and “You can’t prompt this” he says. In January 2026, Almond Breeze released a video featuring the Jonas Brothers titled 'No AI Needed.“ Brands like Le Creuset began explaining their production processes through pinned comments on their Instagram posts.

So in 2026, production quality should not just mean ’perfecting“. A balance of real surface texture, product scale, physical light response, material character and controlled but believable retouch must be maintained. Overly plastic aesthetics that make the real photo look like AI can undermine a brand's trust capital.

This trend is also influencing the aesthetic aspect of commercial photography. A distinct trend is emerging in fashion and lifestyle photography: less polished, more dynamic, and less flawless images. Direct flash, slight grain, unplanned details within the frame, and the clutter of real-life spaces. The reason isn’t nostalgia; The flaw is now an authentication signal It is read as.

However, it would be a mistake to directly apply this trend to product photography. In e-commerce, the product’s color, texture, and scale must still be clearly discernible. The right balance is this: The product information should be flawless; the presentation should have flaws. In other words, the product itself must be depicted accurately from a technical standpoint; but the scene, lighting, and composition must make the viewer feel as though they are actually present in a physical space. This is the balance that AI struggles with the most— because the models tend to converge toward an average aesthetic, that is, toward perfection.