Milanese by birth, Californian by adoption (he teaches at the University of California, Berkeley) Alberto Sangiovanni Vincentelli, an Alumnus of the Politecnico, is among the world's leading experts in computer science. However, it would be mistaken to believe that his interests and knowledge only cover this area, as he is also a great lover of philosophy, literature and economics. “In my life I had to choose which roads not to take, rather than which ones to take”, he told us during a long conversation between Italy and California.
The occasion of the interview is yet another award he has received for Computer Science, the prestigious BBVA Foundation's “Frontiers of Knowledge Award”, which is a sort of lifetime achievement award, similar to the Nobel in both the procedure and the way the victory is announced. "They called me in the middle of the night!" says Vincentelli. This very prestigious awards is assigned for eight different sectors: Vincentelli's one is Information and Communication Technologies (ICT).
In the words of the Committee’s citation, Vincentelli was assigned the award for his “radically transforming the design of the chips that power today’s electronic devices, giving rise to the modern semiconductor industry”. How did he do that? By creating new Electronic Design Automation (EDA) tools, which means he devised algorithms and programmes that optimise the design of integrated circuits (known as chips).
Let us go into detail to understand more. "In chips there are many elements, called transistors, that have to be placed on a planar structure", Vincentelli explains. "These transistors must then be connected to each other via electrical wires which carry a delay: the goals, therefore, are to make connections as fast as possible and to use as little space as possible to accommodate transistors and interconnections”. Once the transistors are in place, one must ensure that the chip works. To do this, there are simulation tools, which are used prior to manufacture: "Today we call them digital twins, i.e., mathematical models that digitally simulate what will happen physically to let us know whether the system works", Vincentelli explains. "Simulation was one of the first things I was involved in, in the early 1970s: it made it possible to speed up projects a lot. Another milestone in the development of the semiconductor industry was the automation of the production of the photolithography masks used to direct beams of light onto the chip to create connections and transistors. In the early days of this industry, in the early 1970s, the masks cut out by hand, but at the end of the same decade, the exponential increase in the number of transistors theorised by Moore's law made it no longer physically possible to do so. Over the years, EDA has enabled an individual chip to shift from having a few hundred of transistors to having millions”.
In his life, Professor Vincentelli has done so many things that a book would not be enough to list them all. In the next hour, our conversation is an immersion in the past amidst establishment of EDA companies - such as Cadence Design Systems and Synopyis, which together are listed on the Nasdaq at around 120 billion dollars -, collaborations with the likes of BMW, General Motors and Magneti Marelli, and success in the academia. While he was teaching at the Politecnico under a contract equivalent to today’s Associate Professor’s, he embarked on a semester-long experience at the University of California at Berkeley, encouraged by his senior colleagues from the Poli. He was young at that time, only 27. At the and of this experience, the request comes for him to remain in the USA. He was hesitating (in Milan I had a great contract, my life, my friends, he explains), but Berkeley did not let up, and after an official request and the granting of permission by the Politecnico, he left for California. And there he remained.
We are in front of guru of ICT, so we take the opportunity to ask him what his opinions are on the latest technology: “EDA is constantly improving, but recently, there has been no breakthrough development in the field: the most innovative thing over recent years is the extension of EDA algorithms and methods to other fields, for example, the targeted design of drugs”, Vincentelli explains.
In the field of semiconductor utilisation, a trend is underway whereby companies that have employed chips bought from specialised semiconductor design companies such as Intel, are jumping into the adventure of designing chips optimised for their own purposes. Apple was the first, but Tesla truly revolutionised the car industry with its project. In essence, Tesla "took everyone in the automotive world by surprise: its car is an electronic system dressed up as a car. This is, in my opinion, one of the greatest innovations of recent times: the fact that a car company has designed a chip”. Does this mean that a new revolution is on the horizon which is not only technological but also industrial, with Amazon, Google and Microsoft designing chips optimised for cloud servers?
“Well, of course, there is also Artificial Intelligence: but I'm not really a big fan of it”, he admits. Why?
"First of all, we should understand the difference between Machine Learning and Artificial Intelligence, which are not the same thing: Machine Learning is a part of Artificial Intelligence. It is using mathematical models to approximate, based on a large amount of data, the functioning of a system the functioning of which we do not know or is too complex to describe. The parameters of this model, which usually takes the form of a many-layered neural network (Deep Learning), are many and must be chosen in such a way as to minimise the approximation error (training). This process is very expensive in terms of calculation (among other things, the energy consumed to carry it out is very high...), and in addition, it does not allow us to understand the reason of certain answers. Well, I don't like this: I want to understand!". What Vincentelli just cannot swallow is the failure to find the reason why Machine Learning gives certain answers: "I have to get to the bottom of why things happen, and with a certain kind of Artificial Intelligence I can't do that," he explains. "The introduction of large language models (immense neural networks used to synthesise and process human language) has made headlines everywhere. ChatGPT, introduced by OpenAI, has raised debate at all levels. The ability to interact with humans with natural language is indeed a revolution. However, when ChatGPT is used to answer questions posed by a user, it answers based on information available on the Web. Given the size of the model, it is practically able to summarise any information on the net. But on the Web there is misinformation (fake news), racial hatred messages, texts against disadvantaged groups or against women, and therefore that ChatGPT (or equivalent) has the potential to influence a large number of people in a negative way”. Many people, including Machine Learning pioneers such as Geoff Hinton, have expressed great concern about the future use of technology! And what about Artificial Intelligence becoming aware? Will it be possible, in the future, for a new, artificial 'species' to appear? "I don't think so: sometimes we may have the impression that Artificial Intelligence understands everything and is sentient, but that can't be, it is still the expression of a machine! It’s us who built it.
What are the most important lessons you have received in your career? "The first one was from a professor of mine at the Politecnico, Giuseppe Grandori, who taught construction science: it was he who made me think about the ethics of engineering. He explained us that the calculations made during the construction of a civil engineering building (at least in the 1960s when I attended the Poli) even took into account a possible collapse of the building and estimated related costs: between an inexpensive building that would resist earthquakes in 80% of cases and collapse in 20%, killing a certain number of people, and a stronger but more expensive building the collapse of which would cause fewer dead, the former could be chosen when its overall cost, including compensation for damage, and the probability of collapsing are lower. And GPT, to tie in with what we said before”.
"The second lesson I learnt was when I arrived in the US, where an engineer is considered in the same way as the technician you call to solve your hotel room toilet flushing problems... or for running a train. Just the contrary of countries like France, Japan and Italy, where engineers are at the top of the social ladder. And not only that, in the scientific world engineering was ranked well below 'noble' sciences like physics. In spite of this, I believed and still believe that engineering is a real science, on a par with all others! This conviction comes from my readings of St Augustine and Kant who spoke of the fact that God or the 'real' is outside the concept of time and space, which are human categories. We engineers also create systems from an idea that comes from outside of time and space, and only after that, at the very end of our creative work, we come to face space and time in the physical world”. Are you saying that engineers are God? "Well, at least in the creative part of our work, we are!", Vincentelli laughs.
If you could choose to be the inventor of anything, what would you like to have invented (or designed)? "I would say the mRNA, because it is a real revolution with immense social impacts: it has an incalculable potential in healthcare, now they even talk of a possible anti-cancer vaccine based on mRNA. And also the CRISPR-Cas-9 gene editing technique, which has been perfected at my University, as it will enable us to cure genetic diseases in the future”.
Any regrets? Any remorse? “On the spur of the moment I would say no, everything went too well for me”, Vincentelli admits. "My regrets are more on doors I closed: I would have liked to study philosophy at University, but also economy or physics, and maybe even to be a bit of a scholar. There was a time when I used to write poems!". "Remorse, only one: not being with my family enough. I must say, however, that I have a very good relationship with my two children, from whom I hear and whom I see very often, and my wife has always supported me in difficult times: definitely, I cannot complain”.
