Self-learning AI represented as a closed loop

The term "closed loop" is used for an automatic control system in which a process is regulated by feedback.
An input is compared to a set point.
The difference between the imput and the set point is amplified and filtered in order to deliver an command output.
This output is filtered again and linked to the input (feedback loop).

Auto-learning AI may be represented as multiple closed loops.
The main closed loop tries and keeps the system into safe boundaries and leads it to "satisfaction".
Other closed loops deal with utilities and help managing system resources.