Generative AI has become an ordinary production means of vertical short videos, but it is difficult to judge the effect on users from platform log data only. Does the creation of visual symbols help attract users’ attention more quickly? How are gaze patterns distributed among different Dynamic Areas of Interest (DAOs); Under what circumstances do generated Symbols facilitate perception over inspection? Nine-six short video user groups, age ranging from 18 to 35 years old. The stimulus pool contained 72 clips balanced by production mode, including fully AI-generated, hybrid AI-assisted and human-made videos. Each person saw 18 video clips in random order. Following gaze quality control, there remained 1,641 participants’ videos. Dynamic AOIs are annotated with contents such as faces/characters, objects’ movements, text labels, motion boundaries, call-to-action areas, etc. Time to the first fixational eye movement, symbolic duration reaction, refixation rate, re-entry probability, transition uncertainty, pupillary dilatation, identification corrects, continuous motivation levels for individuals.
