What still tells Humans and Machines apart

A good definition of intelligence, provided among other things by one of the pioneers of Artificial Intelligence (i.e. Demis Hassabis, founder of DeepMind), argues that:

“Intelligence can be thought of as the ability to convert raw information into useful and usable knowledge.”

Implicit in this definition is the ability to learn, a capacity that until not long ago, was considered to be one of the specific prerogatives of human beings.

With the growing success of Machine Learning, particularly in its most recent variant known as Deep Learning, it seems that machines are doomed to undermine the primacy of humans in one of the skills that, more than any other, has always characterized our species.

Even a recent EU project, which intends to recognize machines as subjects of rights and responsibilities, is based on a difference of degree, rather than substance, between human beings and the so-called “intelligent machines”, implicitly assuming that in the near future, even machines can be characterized by those higher functions (such as intentionality and self-awareness) hitherto considered the exclusive prerogative of human beings.

All this raises serious doubts about the appropriateness and verisimilitude of such a scenario, and suggests a more relevant question:

Are all these claims the result of a misunderstanding, due to the inappropriate use of the term “intelligence” when applied in an anthropomorphic way to machines?

Shouldn’t we more appropriately define machine’s skills as ‘mechanical’ or ‘automatic’ learning, reserving the adjective ‘intelligent’ for human beings only?

Aren’t machines effectively simulating only some of the human cognitive abilities, without affecting the more properly human characteristics?

This is what we will try to explore and clarify in this essay.

(To be Continued)