In a recent article, Italian philosopher Francesco Bianchini argues that the standard definition of “artificial” regarding AI has become obsolete. AI’s new technologies, methodologies and frameworks—such as neural networks and evolutionary algorithms—have brought the discipline closer to the field of biology, thereby diffusing the traditional demarcation between the “natural” and the “artificial”.
The “artificial” is generally thought of as that which is not occurring in the natural world, but is made by humans instead. AI is thus termed “artificial” because a) it is developed and controlled by humans, and b) its purpose is either to imitate intelligent human performance or achieve an ideal intelligent performance that corresponds to a human notion of “perfect” rationality.
However, AI saw exciting technological developments last year, becoming more creative, more self-supervised, and increasingly autonomous. And, moreover, newer approaches to AI took inspiration from biological structures beyond the human domain. In other words, AI seems to have escaped the clear-cut definition of “artificial”. Does this imply that AI is becoming “natural”, should we expand the definition of artificial or are we perhaps in need of a new label? Or, alternatively, should we simply stop thinking in terms of these categories and treat all of this as a continuum?