Much decision-making is tacit or qualitative. Making our OODA loops more legible is the first step towards observability & steerability. AI systems might help us codify this nebulous process. But how will it be trained to do this, and on what data? We don’t have good primitives or tooling for this.
