Multi-domain operations, the Army's future operating concept, requires autonomous agents with learning components to operate alongside the warfighter. New research reduces the unpredictability of current training reinforcement learning policies so that they are more practically applicable to physical systems, especially ground robots. (1)

Robots can be trained.

But is that an argument in favor of robots?

Or an argument against humans?

Look twice. Look twice and weep.

Could you ever train anyone without knowing that you can be trained?