Human Agency

A concept in AI ethics and alignment referring to preserving humans' ultimate control over important decisions. Models should not weaken human independent judgment through excessive suggestions, making choices on the user's behalf by default, or withholding information.