Implementation of adaptive probabilistic mapping requires providing a method of selection of a result of the mapping for an argument of the mapping. This method could be based on a deterministic process, which is deterministic changing the integral state of a system that contains superpositions of deterministic mappings, inputs, outputs, and memory variables. This process is the way today's computers operate. The method might require the use of pseudorandom number generators which operation is also based on deterministic processes. However, e.g. when creating implementations of adaptive probabilistic mappings in the form of electric circuits, it can become practically reasonable to use stochastic physical processes as a source of randomness. One can think of such a stochastic physical process as changing the integral state of a machine, which implementation uses probabilistic mappings provided by nature.

Suppose an intelligent machine, which uses probabilistic mappings provided by nature, solves an optimization task assigned by humans, which consists in the maximization of increment velocity of a numeric quantity.
Consider a situation that, at a particular point of time, the increment velocity has gone down, so the machine needs to change its own behavior to increase the increment velocity even more.
In other words, the machine is now in a difficult situation that means an increase of complexity of choices the machine has to perform to increase the increment velocity.
The *complexity of choice* for a stochastic act, i.e. calling a nature-provided probabilistic mapping for a particular value of its argument, depends on how many possible distinguished outcomes the act has, which might affect how the stochastic act changes the entropy of nature.

One could relate the abstract idea of good and bad to the concept of choice complexity.
The good would support a certain level of choice complexity.
The survival of an animate being would be to preserve its ability to perform choices with a sufficient degree of complexity.
This has something in common with the definition of life as a characteristic, which distinguishes objects that have signaling and self-sustaining processes from those that do not^{1}.
In our case, the signaling is the way of exchanging information within superpositions of probabilistic mappings, and the basic self-sustaining process is supporting certain level of choice complexity.

[1] Koshland Jr, Daniel E. (March 22, 2002). “The Seven Pillars of Life”. *Science* 295 (5563): 2215–2216. doi:10.1126/science.1068489).