I'm more worried about unethical applications of current ML SOTA than AGI. A trivial example would be removing humans out of the loop in antipersonnel armed dro

I'm more worried about unethical applications of current ML SOTA than AGI. A trivial example would be removing humans out of the loop in antipersonnel armed drones so they autonomously classify targets and if/when to pull the trigger.