Doctoral research by Antonella Cosola 

Model: Generative image AI + Intervention

Social-psychological testing ➔ Theme: Counterspeech against sexist hate speech 

Online sexist hate speech is a growing problem, and traditional moderation — deleting comments, banning users — often just pushes it elsewhere. This research tests counterspeech: direct responses that challenge hateful comments and blunt their impact. Since writing good counterspeech takes time, skill, and exposure to harmful content, the study simulates an AI agent modelling different counterspeech strategies in realistic social-media scenarios, to see whether large language models can help at scale. The open question: what makes people willing to speak up when AI acts alongside them, without it replacing their own responsibility to act. 

"People tend to trust an AI's moral judgment less than a human's — so poorly designed AI collaboration risks creating dependency rather than empowering people to respond themselves. "

Antonella Cosola , Uninettuno 

Supervisors: Alessandro Pollini (Uninettuno), Marinella Paciello (Sapienza), Francesca D'Errico (Università di Bari)