Visual Persuasion (VisPer)

What is VisPer?

Persuasive visual communication is becoming increasingly central to modern AI systems, ranging from recommender systems and advertising platforms to multimodal AI agents and generative assistants. These systems not only interpret visual information, but also generate multimodal content capable of influencing beliefs, decisions, emotions, and behaviors.

Despite rapid advances in multimodal foundation models, existing research primarily evaluates systems based on recognition accuracy, reasoning capability, or generation quality, while largely overlooking persuasive impact and communicative intent. In practice, persuasion often operates implicitly through symbolism, metaphorical structures, emotional framing, contextual cues, and cultural references that require sophisticated multimodal reasoning.

The VisPer workshop aims to foster interdisciplinary discussion around persuasive visual reasoning and generation across computer vision, multimodal AI, generative modeling, human-centered AI, computational social science, and creativity research.

In particular, the workshop focuses on several emerging challenges in persuasive multimodal AI, including:

• understanding implicit communicative intent in images,

• modeling creativity and persuasive strategies,

• evaluating persuasive effectiveness,

• cross-cultural and personalized persuasion,

• and developing responsible and fair persuasive AI systems.

By bringing together researchers from academia and industry, VisPer aims to establish visual persuasion as an important new research direction in multimodal AI and computer vision.