AI-Generated Advertising: Consumer Trust and Purchase Intention in Bangladesh
Year: 2026
Type: preprint
Study design: Observational Study
Abstract: Background: Marketing through social media is turning out to be a significant part of the economy for Bangladesh, as one of the fastest growers in economy in one of the largest markets of the world; South Asia. Bangladeshi social media-based marketing adapted AI-generated content as a marketing tool-kit, this study aims to measure the impact it has on the perspectives of the consumers as it grows as a popular choice amongst marketeers. The sentiment was measured through surveys and observing real sentiments of the comment sections of the ads and try to identify a pattern rather than a generalizable conclusion. Objectives: This study examines three holistic questions that tries to assess this overall situation: can Bangladeshi consumers tell AI-generated advertising apart from conventional work; does that recognition alter trust and purchase intention; and which internal factors (trust, perceived risk, aesthetic judgment) plus external factors (language, income, platform behavior) explain any of the difference observed. Methods: Two complementary datasets were gathered for the project. The first comprises 563 unsolicited comments taken from the public social-media pages of 25 Bangladeshi brands. The second is a survey of 131 respondents that captured AI detectability, trust and purchase intention on seven-point Likert scales. Comments received manual coding for sentiment, theme and language or dialect. Survey answers were summarized with descriptive statistics and then cross-tabulated by gender, age and income. Results: Participants place their own detection skill at 5.02 out of 7, yet assign AI-generated content a trust score of only 2.85—almost identical to the 2.82 baseline they give ordinary social-media brand claims. Average purchase intention for social-media advertising sits at 3.92 and drops to 3.30 once respondents believe the advertisement was produced by AI, a reduction of 0.62 points or 15.8 percent (p < .01). Inside the comment corpus only 2.8 percent of posts name AI directly, but 62.5 percent of those posts are negative. The negative reactions cluster among code-mixed Bangla-English commenters and among mid-to-upper-income respondents. Conclusions: These patterns underwrite a four-principle, risk-tiered framework that brands can use when they deploy generative AI in advertising creative.
Source: Preprints.org
Language: English
Subfield: Marketing
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7203701290