Risk avoidance is where one deiberately adopts a strtegy hat reduces risk, for example prefering solutions in a gam space that have lower expected payback, but less variability. For exampel, the minimax algotihm is effectively about minimising the worst case loss against a perfect player. A less risk averse approach against an imperfect opponent might be to choose a path that has a small potential to lead to a bad defeat, but has also many ways once can turn the game around and win. Stochastic games where there is an element of chance may mean an AI player has to choose between options that have low potential yield, but also low uncertainty, and those with higher expected gains,but with a lot of uncertainty. By adjusting the balance between these one can make the asystem more or less risk averse.
Used on page 236