TECH | Mar 5, 2019

5 lessons from the book “Le macchine sapienti” (The wise machines)

Between artificial intelligences and human decisions. The new book by Paolo Benanti leads us to reflect on the effects of machines making decisions

What is happening with the implementation of new forms of Artificial Intelligence and with their ever-increasing diffusion? What are the considerations to be made and the risks to be evaluated?

The book “Le macchine sapienti ” (The wise machines) by Paolo Benanti, specialized in bioengineering and neuroscience, takes readers on a journey through the awareness of what is already reality and of what will happen in the near future, with two interludes on the ethical question and on how machines will have to interact with man.

What are the 5 things in the book which remain once page 156 has been turned?

1.We are surrounded (by artificial intelligences)

With a long series of practical examples, divided by theme, the first pages of the book show how much “artificial intelligences are pervasive and are weaseling their way into every area of our existence. Both in production systems, becoming robots, and in management systems by replacing servers and analysts”. But also and above all in everyday life, where our smartphone ends up becoming “an actual partner which interacts in a cognitive way with the user”.

2. Algorithms risk giving rise to social injustices or discriminations

One of the considerations which strongly emerges from the pages refers to the possibility that “algorithms can bear “prejudices” – calculation-based, not moral-based ones – which applied on a large scale risk giving rise to social injustices or discriminations”. Here then, the need to reflect on innovation management. Because “only an innovation which becomes authentic human development will create a better future, avoiding dystopian nightmares”.

3. For the first time, machines are the true players in the transformation, not men

If it is true that Artificial Intelligence is already so pervasive, we should be aware of the fact that for the first time in history it will not be a human elite which drives a far-reaching change. And this aspect “must be investigated in all its complexity. To understand the challenges and potentialities of artificial intelligences, we must first understand the human side of the human-machine relationship. We must start from the foundation, from the specificity of human cognition in order to understand what its own contribution is and what that of artificial intelligences, and how these handle what falls within their jurisdiction”.

4. Data alone cannot guide our choices

We know that “wise” machines act through data analysis, on the basis of which they “grow” and take autonomous decisions. But do data alone drive people’s decisions? To answer this question, there is a simile in the book: “Data may be understood as a map of reality. A datum is an elementary description, often codified, of an entity, a phenomenon, a transaction, an event or other. As coding, a datum is a reduction of reality similar to that of a geographical map”. That said, can a data map be considered a copy of reality? The answer is no, and because “machines make decisions based on maps which are reductions of reality and by using data from sensors which perform a similar reduction, they have a fallibility factor”. Ethics, therefore, must guarantee a fail-safe system or at least a degree of fallibility which does not harm those values which are considered key and fundamental.

5. A governance of artificial intelligences is urgently required

An entire chapter is dedicated to the problem of AI governance, considered “the instrument with which to ensure that this synthetic cognition, made possible by technological innovation, does not come to take on dehumanizing forms. Governance is the space where anthropological and ethical observations must become effective forces and an organizational culture in order to shape and guide technological innovation, making it an authentic source of human development”.

Sonia Montegiove