Is machine learning open-ended? – Towards Data Science – Medium

In biological theory, there is only one creative'force' that is capable of yielding open-ended evolution -- natural selection (NS). Although Darwin's original description of NS was with explicit reference to features of modern biology like organismality, generations, sex and so on, there has been a growing movement of Universal Darwinism that has come to see NS as substrate neutral (e.g. As such, NS can be thought of as an algorithm for performance optimisation. In this way, the mathematics that is used to describe NS can equivalently model true learning (Plotkin 1995). Although NS is typically used in reference to evolutionary change that occurs in genes between generations, selection can also describe the learning process that occurs in memes within an individual.