Why Am I Set at 68?

Chris Argyris used a thermostat to explain two kinds of learning. A thermostat set to 68 degrees turns on the heat whenever the room drops below that. It detects a gap and corrects it. That is single-loop learning. A thermostat that could ask "why am I set at 68 degrees?" and look into whether another setting would heat the room more economically would be doing double-loop learning.

Most organizations are excellent at the first loop. A missed target triggers a review, the review finds a process error, and the process gets fixed. The target itself is rarely on the table.

AI makes the first loop dramatically faster. It will find the gap, propose the correction, and run the next cycle before the meeting ends. It will not ask why the setting is 68, because the setting is the one thing it was given.

This is where meta-learning earns its name. It is the habit of examining how you reached a judgment, including the goal, the metric, and the assumptions that once made the setting look sensible. Argyris also observed that highly successful professionals are often the weakest at it. They have had little practice at failing, so defensiveness arrives exactly when learning is needed.

A useful test for any AI project: after the first quarter, what did the team change about the goal, rather than the method?

What setting in your organization has not been questioned since someone first chose it?