PEOPLE | Dec 28, 2017

Data, AI and Machine Learning: interview with Mauro Campanella of GARR

What is, and what will be, the role of so-called "automation" in the research sector (and elsewhere)? How should it be managed?

“Historically, if we think for example of the development of communication protocols for the Internet, the research networks environment has created or anticipated technologies: this is why we can say that research is at the forefront. Mauro Campanella, head of research and international development at GARR, the ultra-broadband Italian network dedicated to the education and research community, explains the role of the consortium in this way with respect to innovation.

“We are also currently carrying out research on artificial intelligence e machine learning, activities that I would quite simply define as necessary. Necessary, where thanks to automation and machine learning, you can avoid people having to perform repetitive tasks and utilize these technologies to amplify their range of activity, as in our case. What matters  – continues Campanella – is that the term “intelligence”, in reference to machines is never confused with the term intelligence referring to man, because we are talking about two completely different things.

How true or false is the perception that we have today with regards to the possible link between loss of work and robotization?

“I would make a distinction between the potentials of technology and the use of technology. Humans are very quickly developing the possibility to transmit (Internet), produce and process information (powerful, miniaturized computers) and store it (silicon memories); always in large quantities. On this basis, very complex software, which only a few years ago was too sophisticated to be in common use, can work effectively. This is without considering the developments that have also occurred on the hardware and computing capacity side, as well as data storage and transport. One example is the mobile phone in our pockets, which is now also a hugely powerful calculator. In my experience these tools are changing the way every task is being performed and opening up new possibilities, some of which are very positive, such as communication, research and education. Moreover, due to their digital nature, these new instruments are suitable for use in every human environment and are particularly successful in those that can be automated: from driving, to medicine, to forensic practice. Clearly, as with any new tool, or human action, ambiguous uses are possible, just as changes are likely. Therefore, in the relationship between human beings and technology, we are facing a radical change that influences our lives and culture. Humans will increasingly be flanked by technology. Regulating it without inhibiting progress and innovation in fields such as medicine is the challenge.

On a final note, artificial intelligence, in my opinion, has really developed since we gave up the will to create it with algorithms and we let the systems evolve via “self-learning”. The new machine learning technology is therefore particularly innovative: it radically modifies the traditional approach based on man’s initial description of the task, since this is replaced by the machine itself learning (for example through “silicon networks”) and creating, with or without man’s control of the method to achieve a goal.

It is a great opportunity, of course and one which we must continue to monitor when these technologies are applied to areas where human intelligence is indispensable. If we consider, for example, the military sector, or that of weapons in general, we realize the possible dangers and the fact that it is probably too soon to “automate” in this sector. This is why great caution and the development of an ethic are needed in the use of data, technology and above all results.”

Which sectors, more than any others, will be completely transformed in the future? How should we prepare for this?

“Probably all the sectors that can be automated will see progressive and constant changes. Not only industrial automation, but also that of the tertiary sector, such as banks that are increasingly replacing people at the counter, sometimes not taking into account their users’ need for digital skills, who thus risk being excluded, such as senior citizens.
Preparation will not be easy: I’m thinking of a group of skills as tasks change, updating to new technologies and above all I’m thinking of education, which will have to change from mnemonic training to the ability to manage information and constantly update skills. Our education systems will have to adapt and they will have to do it quickly: if we think of the number of students enrolled in technical institutes in Germany compared to us, we realize how much we still have to do from this point of view.”

 How important is data culture today?

“The initial data is now constantly monetized by companies that work on the net (for example the majors such as Google, Amazon, Facebook and others) and the intrinsic value in information is increased by its analysis, especially when it is linked and cross-linked in large quantities.
The value of the data culture increases proportionately to its being accompanied by an understanding from the scientific method that takes account of source clarity and methodologies, analyses and rigorous treatment. In this sense, I have the impression that beyond the narrow fields of science and the industry of excellence, the rigorous analysis of data and a culture focused on its accuracy and centrality may not be sufficiently widespread and shared in Italy.
It is not just there for economic uses; there are examples of application of the analysis of large quantities of data provided for example to art that has offered both new culture (as in the case of works seen at various wavelengths) and a new understanding of art (I think of The Last Supper digitally revised by Greenaway) and they have demonstrated its value.

Apart from scientific research, which has always based its processes on data, sensitivity with respect to the importance of data has developed unevenly, with some companies beginning to use information extensively and others that are still a long way off. If we think of loyalty cards, for example, we understand how much the data collected through these (certainly not from today) can be a source of valuable information that ends up “monitoring” not only the supermarket that sells but also the companies that work with them. Companies must become aware of the fact that data forms the basis of automation and optimizing processes and will increasingly have a fundamental effect on choices, also reducing the time needed.”

Which GARR projects in the pipeline do you consider the most interesting?

“Preparing for change is essential for GARR to serve the demands of research and education.  We have started an interesting project called ELISA, capable of rethinking infrastructures and services, using precisely those technologies that today enable automation and machine learning. If we think of network devices, to give an example, we realize that from the hardware point of view, they have a common structure. And this leads us to say that in many cases we could use similar components and software that virtualizes them, to have more modularity, flexibility, and adaptability to new needs. Automation, moreover, presents this as a primary advantage: the ability to quickly change processes and services, since updating software can take less time and resources than having to “change” machines or completely update human beings. The aim we have at GARR, and that all innovative companies have, is precisely that of offering more with even fewer resources.
In “transforming” our network, I like to underline the fact that we envisage close collaboration with users and other European networks, to offer the same possibilities to all users, regardless of where they are.”


Sonia Montegiove