Machine learning and learning machine


The HR Director of the World Economic Forum explains how to move away from an out-of-date educational learning system


Machine learning and learning machine

By the time you have read the first three lines of this article, the Visa network will have processed about 10 million transactions. Just now, Google is analysing 56.000 searches which, in turn, will provide billions of results. Right at this moment, 3 million emails are sent, most of which from systems and not people.

The American government created a research centre called BRAIN, “Brain Research through Advancing Innovative Neurotechnology initiative”, and invested on it an astonishing amount of money. Elon Musk, the ingenious founder of Tesla, invented Neuralink, which aims to create a connection between our brain and computers: in other words, the target is to “download” our brain. McKinsey estimated that in 2015 20 to 30 billion dollars were invested in Artificial Intelligence (AI).

Artificial Intelligence
The specialists working on AI – who, incidentally, are still a few and earn astonishing wages – agree that, around 2030, the so-called technological singularity will occur, namely the moment in which machine learning will overcome human learning. How can this be possible? It has been estimated that an adult human brain has around one hundred billion neurons, which are chemically and electronically connected one hundred trillion times. Making a quick calculation, the human brain remains the most complex machine on earth.

Knowing that AI will overcome human intelligence in about a decade makes people anxious, to say the least. Have we lost already? What can we do about it? In this article, I don’t envisage a battle between man and computer, a subject which has been much debated for decades. I would like to illustrate how we can move away from an out-of-date learning paradigm to a new different one both as individuals and as a society. In other words, as regards AI and machine learning, namely the computer ability to learn without having being programmed, and deep learning (a science which has been rapidly developing), it is down to us to rethink what we mean with learning.

In the old paradigm, three stages were identified in the span of a human life: a studying and learning stage up to 20 years-old, a working stage for the following 30 to 40 years and a retiring age (if we are lucky to have one and are still alive and in good health). All societies were structured on the grounds of this paradigm. What has changed?

Life expectations
We have moved from a life expectancy of about 40 years at the beginning of the Nineteenth century to a current one of 75 years, and are bound to reach an expectancy of 90 to 96 years in 2050, when half billion peoples will be over their nineties and over 50% of the world population will be under 27 years of age.

Retirement plans
Unsustainable. At the moment, in Europe there are 4 workers for a single pensioner; in 2040, there will be only 2 workers for a pensioner. The outcome is self-evident.

Labour market
Unemployment rates are bound to increase among people holding low or middle low education, while few “privileged” categories will prevail such as experts on AI, computer science, robotics, nanotechnologies –professions generated by the fourth industrial revolution – and scientists because of their winning skills.

Industrial revolution
Which is the main difference with the previous three revolutions? The speed of the last revolution, which is staggering, not linear and the fact that it is not only changing products, services and costs but also us and the relationships among humans and between humans and AI. What does this mean to us? We cannot rely on the old paradigm any more, an out-of-date system connected with the earlier industrial revolutions. The new reality which is rapidly emerging is the reality in which we will live for 90 to 100 years and in which we will have to keep learning throughout our life.

In other words, to balance machine learning we need to become ourselves learning machines. We cannot allow ourselves to simply “unplug” but need to keep our mind active. Be careful, I don’t mean being connected with social media 24 hours a day, an activity that reduces our attention and critical thinking. I mean developing and keeping our brain trained, motivated by an endless curiosity and a will to improve ourselves.

The winning skills are the human-related and learning-based ones. The ability to collaborate, to understand complexity and adapt, to have a global vision and not a faceted vision due to overspecialization, to develop mutual understanding and build trust, all skills that AI will never be able to achieve. In other words – in my opinion – we, as human beings, are going to lose a speed competition but we have good chances to win deep-learning and mutual understanding competitions. A redesign of school and university curricula is certainly required: for example, Finland has radically changed the way school subjects are told since primary school. There are no longer geography, history or maths classes; the students study the Second World War taking into account geopolitics, economy, history and geography. So, they learn how to cope with complexity and how “to put pieces together” developing a thinking system.

Thus, in order to learn how to be human, we should never stop being curious and eager to learn and avoid becoming overspecialised, being aware that failure is a learning process – like the saying goes “you learn by making mistakes”. In the Sistine Chapel, one of the best-known art work in the world, a scene portrays God giving life to man. Few people have noticed that Michelangelo painted God inside a frame that looks like a human brain. Michelangelo told us that God gave us life and knowledge. Today more than any time before we need to use it, relying on the stronghold of our ethics and holding the seat belt fastened.

L'autore

Paolo Gallo

Paolo Gallo