1 paper
Ondrej Bajgar, Peter Tisnikar, Alessandro Abate +2
The development of safe and beneficial AI requires that systems can learn and act in accordance with human preferences. However, explicitly specifying these preferences by hand is…