AI cannot deescalate situations – opponents don’t know How to surrender to a robot, and the robot will see that as threatening.
Docherty, 2014 - senior researcher in the Arms Division of Human Rights Watch [Bonnie “Shaking the Foundations The Human Rights Implications of Killer Robots” Human Rights Watch http://www.hrw.org/sites/default/files/reports/ arms0514_ForUpload_0.pdf Acc 12/27/20 TA]
In addition, the deployment of fully autonomous weapons in law enforcement situations could affect the actions of the individual posing a potential threat. He or she might not know how to behave when confronted with a machine rather than a human law enforcement officer. The individual might respond differently to a robot than to a human and as a result unintentionally appear threatening. A robot’s misinterpretation of the necessity of force could trigger an arbitrary killing in violation of the right to life.