INTRODUCTION: Image-based affect induction research is constrained by weak to moderate emotional elicitation effects, limited stimulus diversity, and minimal cultural tailoring. Advances in generative AI offer new opportunities to produce scalable, customizable, and context-aware affective stimuli.

AIM(S): We developed a human-in-the-loop pipeline for generating context-aware affect induction images and established and validated the initial Library of AI-Generated Affective Images (LAI-GAI), with stimuli tailored to cultural, sex, and age contexts.

METHOD(S): Using generative AI guided by established emotion taxonomies and existing affective datasets, we produced 847 images and corresponding descriptions spanning 12 discrete emotions. Images were iteratively refined in collaboration with local cultural experts. Validation was conducted across six studies (total N = 2,470; participants from 58 countries). Participants rated five image categories: (1) images from existing affective databases, (2) AI-generated images without cultural adjustment, (3) culturally adjusted AI-generated images, (4) sex-adjusted variants (male, female), and (5) age-adjusted variants (childhood, adulthood, older age).

RESULTS: AI-generated images were comparable to established database images in eliciting affective responses. Culturally adjusted images showed greater effectiveness in targeting intended emotions than unadjusted AI images. Sex- and age-adjusted variants produced affective responses comparable to their base images, indicating controllability without diminished emotional impact. We also estimated the smallest subjectively experienced difference for affect induction research (Cohen’s d = 0.05–0.29).

CONCLUSIONS: The LAI-GAI demonstrates that high-quality, scalable, and context-sensitive affect induction stimuli can be generated cost-effectively using a human-in-the-loop AI pipeline. This approach overcomes longstanding limitations in affective research and establishes a foundation for future AI-driven methodologies in affective science.

FINANCIAL SUPPORT: National Science Centre in Poland (UMO-2020/39/B/HS6/00685; 2023/49/B/HS5/02139) and Excellence Initiative—Research University (ID-UB) program at Adam Mickiewicz University, Poznan (140/04/POB5/0001, 151/12/POB5/0005, 174/12/POB5/0001, 177/02/UAM/0012, 181/13/SNS/0003, 198/12/POB5/0007) supported preparing this article with a research grant. The funders had no role in study design, data collection, analysis, publishing decisions, or manuscript preparation.