Where Operators Draw the Line: Trust, Not Capability, Sets AI's Limits

Operators are drawing delegation boundaries around AI based on trust and human stakes, not technical capability, and many lines remain unsettled.

Senior operators worked through which decisions should stay human-led versus delegated to AI. The clearest boundary was people work. Interpersonal relationships, career development, and performance management were treated as firmly human-led. Crisis communications and marketing also required human oversight, often in a human-plus-AI combination with governance. The most striking pattern was that operators justified keeping mass transport and aviation human-led on grounds of consumer confidence, not capability. Trust and perception, not whether the technology works, set the real adoption ceiling. This showed again when operators observed that autonomous transport is more adopted in early-adopter countries than in the US, where deployment exists but public uptake is thin. Geography and culture pace adoption more than engineering. Operators also noticed AI quietly entering training and development through generated videos and tools, and admitted they had not decided whether that work should ever be fully automated. On cybersecurity, positions were unsettled, split between human control and significant guardrails with governance plans. Notably, the group ran out of time before reaching the question of how governance can enable innovation without slowing adoption, which itself signals that boundary-setting consumes more attention than enablement. For executives, the message is practical. Decide delegation lines deliberately, separate trust from capability in adoption planning, and inventory where automation has already crept into functions before a policy exists to govern it.