Intervention1: Testing Consumers’ Emotional Acceptance of AI-generated visual content 

Research Question

How can fashion brands ensure that AI-generated visual content preserves, rather than undermines, consumers’ emotional connection to the brand?

理由

This first intervention aimed to test whether consumers’ emotional acceptance of AI-generated advertising could shift rapidly when exposed to specific visual elements. Prior research suggests that consumer acceptance of AI advertising is shaped by the verisimilitude (realism) and vitality of the visuals, whereas excessive synthesis tends to evoke a sense of eeriness, reducing acceptance (Gu et al., 2024). Based on this, I designed an intervention focusing on realism as a key factor.

All of this stems from a deeper issue: When visual creation shifts from “photographic logic” to “generative logic”, who is determining the “reality” of the visuals? Are it pixels, algorithms, or the psychological threshold of the viewers? Therefore, this intervention not only tests “how consumers view”, but also aims to understand “why they view in this way”.

Method

Task: Participants viewed a set of “highly realistic” AI-generated advertising images (see Figure 1 and Figure 2).
Objective: To observe how their attitudes towards AI advertising evolved across three stages:

1. Original opinion – their general perception of AI advertising.
2. Immediate reaction – their response to a specific AI-generated image.
3. Impact on brand connection – whether and how the image influenced their emotional link with a brand.

Figure 1. Promotional visual provided by Lavie Company (Lavie, 2025).

Figure 2. AI-generated visual created by the author for research intervention, using ChatGPT (Li, 2025).

This design intentionally isolated the “visual variable,” allowing the study to distinguish between participants’ pre-existing beliefs about AI and their actual emotional responses when confronted with realistic AI imagery. This separation helped reveal the gap between abstract attitudes and lived reactions.

Findings

Participant	1. General opinion of AI ads	2. Reaction to AI-generated images	3. Impact on brand emotional connection
48M, brand owner	Positive; depends on professional design and logic.	Very realistic, consistent with brand image.( figure 1 )	No impact; “looks real, cannot tell the difference.”
33F, private jet manager	Thinks AI looks fake; tolerates it for trusted brands, dislikes AI models for influencer brands.	Looked very realistic, aligned with brand.( figure 1 )	Some impact due to missing “real-life connection,” but more accepting than before.
40M, artist	Sees AI as “too perfect, mechanical, lacking warmth.”	Could not distinguish AI from real photos; perception shifted.( figure 1 )	Still prefers human warmth and emotional resonance.
33F, hospital executive	Indifferent for trusted brands; rejects only if it feels “too fake.”	Surprised that AI could look so real.( figure 1 )	If realistic, no impact on brand emotions.
9F, student	Finds food ads unappealing if models look fake; prefers real humans for appetite and safety.	AI looked realistic, less scary, more willing to try.( figure 2 )	Acceptable if realistic, since she “can’t tell the difference.”
37F, art professional	Initially disliked AI as “cold and unreliable.”	Realistic AI looked more natural than over-retouched human ads.( figure 1 )	If realistic and creative, could even improve brand evaluation.
25F, designer	Negative; stressed brands need originality.	Realistic, but still lacked “human warmth and thought.”( figure 1 )	Depends on which parts were created by AI and whether designers were involved.

Across participants, a recurring pattern emerged: while conceptual resistance to AI was common, direct exposure to highly realistic visuals often softened these attitudes. This suggests that emotional acceptance may be more malleable—and more situational—than existing survey data implies.

Key Insights

1. Realism reduces resistance – When AI visuals appeared sufficiently realistic, most participants shifted from rejection to greater acceptance, with some even finding them more natural than overly retouched ads.
2. Brand trust acts as a buffer – Consumers were more tolerant of AI-generated visuals when associated with trusted brands. For unfamiliar or influencer-driven brands, vitality and authenticity were crucial.
3. Lack of warmth – Many participants highlighted that AI-generated visuals felt “too perfect” and lacked human creativity and resonance.
4. Context matters – Realism was most critical in food advertising, directly influencing appetite and perceived safety, while fashion advertising emphasised style alignment and warmth.
5. Generational differences – Younger participants (e.g., children) were drawn to animated AI visuals but ultimately still considered human-based ads more emotionally powerful.
6. Role of designers – Some participants stressed that completely replacing human creativity with AI risks weakening originality and brand trust.
7. Emotional thresholds still favour the human touch – Even when realism was achieved, participants instinctively searched for human intention, imperfection, and relational cues. This implies that emotional connection relies not only on how real an image looks, but on whether it feels alive, intentional, and human in origin.

Preliminary Conclusions

• Verisimilitude (realism) is a key factor in reducing consumer resistance to AI advertising.
• Brand trust and familiarity influence how tolerant consumers are towards AI-generated visuals.
• Warmth and imperfection remain weak points of AI visuals; even highly realistic images may lack “human touch.”

摘要：研究结果表明，消费者对人工智能生成的品牌视觉形象的态度会根据特定因素迅速转变。如果品牌遵循某些心理学原理，就可以利用人工智能工具提高效率，同时最大限度地减少对消费者与品牌情感联系的负面影响。

References

• Gu, C., Jia, S., Lai, J., Chen, R. and Chang, X. (2024) ‘Exploring Consumer Acceptance of AI-Generated Advertisements: From the Perspectives of Perceived Eeriness and Perceived Intelligence’, Journal of Theoretical and Applied Electronic Commerce Research, 19(3), pp. 2218–2238. Available at: https://www.mdpi.com/0718-1876/19/3/108 (Accessed: 14 August 2025).