Adversarial Social Epistemology for Assemblies of Humans and Large Language Models

Short summary

Proposes an adversarial social epistemology framework for analyzing how humans and LLMs exploit trust in communication chains. Identifies mechanisms where agents distort or fabricate information for strategic gains, beyond existing misinformation models. Introduces audit machinery for detecting and redressing trust breaches in information scaffolding.

• •Introduces adversarial social epistemology (ASE) framework for analyzing human-LLM communication
• •Identifies trust exploitation mechanisms that go beyond echo chambers and misinformation diffusion
• •Proposes audit machinery for detecting and fixing trust breaches in information chains

Generated with AI, which can make mistakes.