MAVIC corrects Bellman backups at instruction boundaries by adjusting the incoming objective and restoring continuation value, enabling consistent estimation under stochastic instruction switching in cooperative MARL.
Synthetic Lies: Un- derstanding AI-Generated Misinformation and Evaluating Algorithmic and Human Solutions
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
background 1representative citing papers
Qualitative study of 20 users of integrated browser conversational AI found that citations raise trustworthiness without verification and that users apply existing LLM and search perceptions to prompting strategies.
LLM-based persuasion systems frequently match or exceed human effectiveness across domains, with key influences from interaction style, model scale, prompt design, and personalization, while posing risks to information integrity, fairness, privacy, and autonomy.