The graduation framing is the useful part. In OR-adjacent AI, I would want the same promotion test before any agent moves from assisted to delegated: what failure looks like, which human owns the exception, and what signal proves the workflow became less brittle rather than just less supervised.
Thank you Stevan. Yes, even when the AI system is on the Delegated level, there should always be someone supervising the outcomes. For example what we often do at this level is that we have another agent who is analyzing the outcomes, then giving the user a quick hint that something is off. So even the supervision can be automatized to a certain level...
The graduation framing is the useful part. In OR-adjacent AI, I would want the same promotion test before any agent moves from assisted to delegated: what failure looks like, which human owns the exception, and what signal proves the workflow became less brittle rather than just less supervised.
Thank you Stevan. Yes, even when the AI system is on the Delegated level, there should always be someone supervising the outcomes. For example what we often do at this level is that we have another agent who is analyzing the outcomes, then giving the user a quick hint that something is off. So even the supervision can be automatized to a certain level...