Synthetic intelligence is not really human, so we should always stay away from conditions like “reliable AI” that not only humanize AI but will also indicate a standard of dependability that simply just will not exist. Robustness: AI-powered units need to be actively defended from