Let’s remember ai is software. It has no conscience and therefore cannot judge whether it lies or not.
This raises major issues for ‘black-box’ ai as the algorithms are hidden so the ai’s sense-making for decisioning can start to deteriorate subtlety way before being noticed by most humans. ai black box, unless constrained in its learning, will eventually lead to decision distortions and decision contamination. This raises major concerns about trust, ethics and morals. It does not bode well for Multi-Robot Pursuit Systems (refer to AI 122) though you can hear those in favour ‘its for the greater good’!
Maybe our work on human-controlled artificial intelligence (white-box) will become recognised as these black-box ai issues become more prevalent.
