AGI Behaviorology and the External Boundary Observation Framework

Current mainstream AGI safety research mainly relies on internal alignment, post‑hoc governance and fragmented behavioural statistics. These paradigms suffer from structural blind spots: they cannot reliably trace the continuous evolutionary process before full‑blown AGI emerges. This paper establishes AGI Behaviorology — an observation paradigm built upon external behavioural baselines instead of internal mechanism interpretation. It defines three potential observation anchors derived from human‑AI causal linkage, non‑transferable subjective experience, and physical‑embodiment‑triggered quasi‑experience. A three‑tier monitoring hierarchy is proposed to track dynamic shifts during pre‑emergence, transitional and post‑emergence phases. The framework aims to provide a falsifiable foundation for early‑warning surveillance and prospective safety governance.