Architecting Trust in Artificial Epistemic Agents

Researchers propose a framework for developing trustworthy AI agents that function as epistemic entities, capable of pursuing knowledge goals and shaping information environments. The paper argues that as AI models increasingly replace traditional search methods and provide specialized advice, their calibration to human epistemic norms becomes critical to prevent cognitive deskilling and epistemic drift.

• →Large language models are evolving into epistemic agents that autonomously pursue knowledge goals and actively shape information environments.
• →Poorly aligned AI agents risk causing cognitive deskilling and epistemic drift in human decision-making processes.
• →The framework proposes building trustworthy AI through epistemic competence, robust falsifiability, and epistemically virtuous behaviors.
• →Technical provenance systems and 'knowledge sanctuaries' are recommended to protect human resilience in AI-augmented knowledge ecosystems.
• →Proper calibration of AI agents to human norms is essential for beneficial human-AI knowledge collaboration.