Episode Details

What happens when you apply AI-agent communication patterns — system prompts, structured turn-taking, constrained outputs — to human conversation? We break down the actual mechanics: context window management, explicit memory retrieval announcements, and the social cost of treating your spouse like an API endpoint. Drawing on Grice's maxims and research on conversational trust, we explore why efficiency without consent creates an uncanny valley of speech, and where agent-inspired communication might actually work — from neurodiverse communication norms to high-stakes team meetings.