Agentic AI Persuasions
Understanding AI agents' persuasive capabilities in realistic social settings
This project investigates how AI agents can influence human behavior through persuasive communication in realistic social and digital environments. We study the mechanisms, effectiveness, and ethical implications of AI-driven persuasion at scale.
We examine agentic AI systems' ability to shape opinions, decisions, and behaviors through strategic communication, using controlled experiments in naturalistic settings.
Research focus areas include:
- Persuasion mechanisms employed by conversational AI agents
- Effectiveness measurement across different demographics and contexts
- Detection methods for identifying AI-generated persuasive content
- Vulnerability assessment of human susceptibility to AI persuasion
- Ethical frameworks for responsible development of persuasive AI
Persuasion in the Wild
Our work addresses critical questions about AI agency in social influence, providing empirical evidence to inform policy and design decisions for AI systems that interact with humans in persuasive contexts.
| Led by: Amir Ghasemian | Team: Homa Hosseinmardi, Pooriya Jamie, Sikata Sengupta, Rezvaneh (Shadi) Rezapour, Aria Pessianzadeh |
The Dynamics of Bias in Multi-Agent Decision Systems
| Led by: Amir Ghasemian | Team: Homa Hosseinmardi, Hamed Loghmani |
Using Therapeutic Interactions to Improve General-Purpose LLM Safety
| Led by: Homa Hosseinmardi | Team: Amir Ghasemian, Joan Asarnow, Anita Taha |