Generative AI is rapidly evolving toward increasingly autonomous, agentic systems, raising a fundamental question: how can human judgment remain economically indispensable in AI-mediated work? The Hybrid Intelligence (HI) paradigm proposes a shift from automation-centric architectures toward interaction designs that structurally embed human authorship and decision-making within workflows. This paper investigates how the Frame–Explore–Refine–Commit (FERC) framework model as a metacognitive scaffold for human–AI co-creation, can be operationalized, measured, and integrated into real-world chatbot systems. We present an iterative design science research (DSR) study developing a FERC-based chatbot (“FERC-bot”) across five design cycles. The study demonstrates how adversarial prompting can systematically exploit naive scoring approaches, revealing fundamental limitations of surface-level authorship metrics. We argue that reliable authorship support requires a shift from static output evaluation toward process-aware, interaction-embedded design. We also outline a vision in which FERC-based scoring and interaction principles can be integrated into commercial and custom AI systems, providing both front-end metacognitive guidance and certifiable validation of human authorship. Such capabilities may form a key building block for trust and accountability in a human-touch hybrid intelligence economy.
Accepted for Hybrid Human AI (HHAI) conference 2026.