Giorgio Metta, scientific director of the Italian Institute of Technology in Genoa, was in Rimini at the We Make Future festival to talk about technology transfer, that is, how research comes to generate interesting products for the market, of startups born from the laboratories of Val Polcevera , which aims to become a “robot valley” thanks to one of the “innovation ecosystems” born thanks to funding from the NRP. And of the future applications of robotics and computation made in Italy, which affect factory work as well as prostheses and new biomedical solutions. He proposes a company made by research institutes to break through the market. But the father of the robot iCub, now an icon of robotics and IIT, could not but begin by asking for a comment on the shocking news of a Google employee who raised the alarm: “Artificial intelligence is becoming sentient “.
“Artificial intelligence is becoming sentient”, Google engineer raises the alarm. And Big G suspends it
by Flavio Bini
Doctor Metta, what do you think?
“We were talking about it recently at an event with Luciano Floridi. The question arose because it is a topic that attracts a lot. The answer we gave ourselves is that in Google there are 300,000 employees, and there was one who wanted to get noticed. . I was reading an interview by the Economist with Douglas Hofstadter, a dean of artificial intelligence. He said: “If you have a conversation with one of these new and sophisticated machines, it may even seem that there is something behind it, especially if you keep the discourse on a more or less reasonable path. But have you ever tried to ask one of these software how many hamburgers it takes to make a cumulonimbus? The answers are absurd, the machine does not understand that you are asking it for something that does not mean anything. And it answers anyway “. Behind it there is definitely not a system that understands the question that is being asked. Although they are fantastic and work very well and you can do a lot of interesting things with them.”
Is there this kind of question on the horizon?
“Technically we are not asking ourselves the problem. We are not even working in that direction. The aim we are giving ourselves is to do good engineering, to have things that work to solve problems. But however much someone tries, we are far from one. systematic study, for example, of our brain and cognitive system with the aim of transferring it into algorithms. We see this from the fact that the way things are done typically involves “brute force.” We don’t even worry about how much a machine consumes learn to answer the questions. Our brains consume 30 watts roughly, these computers cost tens of millions of electricity alone. “
Future
The technology of iCub 3, the IIT robot halfway between metaverse and tele-existence
by Emanuele Capone
Will iCub never call her dad?
“The road is quite long and above all it is a complicated problem of both neuroscience and artificial intelligence. It may be that by mistake I can say something like this but surely there is not behind that understanding of the world that we have. It would be necessary to rethink. how we are designing it. There must be an architecture, elements, like that of our brain, and maybe a body. Our mission is to arrive at solutions to solve problems that have meaning for humanity. strategic plan there are three missions: sustainability, health and aging of the population. Which means assistance and help at home, even with robots “.
Let’s talk about the heights of AI and Google. Are we missing an opportunity in Italy?
“I think there is a problem. Perhaps it is not yet lost, and maybe the PNRR will give us a hand to update the infrastructure, have more computing capacity, more people working on this issue … but there is to be very careful, because the competition is very tough and the various Google and Facebook are very fierce. They have large amounts of data at their disposal. But we too, those of the industrial chains, of the production chains, of how things are done. after the incentive of the NRP, if a more serious plan for maintaining resources and investment in some crucial issues such as AI is not made, it could become difficult to keep up “.
Let’s go back to Val Polcevera, what is the innovation ecosystem that has just been approved and how will it become a “robot valley”?
“Val Polcevera is where the IIT was born. In recent years we have expanded towards the sea in the same valley. Here are the Leonardo Labs, Ansaldo Energia, companies that make superconductors and magnetic resonance machines, others with expertise in There is a good ecosystem, we asked ourselves: “Why couldn’t it become a robot valley, together with other companies and our startups?”. We have prepared a project with an important content of technology transfer that has robotics and AI as its main theme, thinking of projects that involve these disciplines and companies. It will be us, the University of Genoa and the CNR, with the Ircs San Martino and Gaslini. It is one of the innovation ecosystems approved with NRP funds, with a budget of 110 million. But there is another project that has just been approved “.
Which?
“We signed a few days ago in Rome with Draghi, the ministers Gelmini, Cingolani and Speranza, and the president Toti, the agreement for a flagship project: the construction of a computational hospital. A total investment of over 400 million. Patients obviously they will be real and we will cure them, it will have 500 beds, and an area entirely dedicated to research: neuroscience, genomics, robotics for rehabilitation, yet to be defined with the executive project. But these are all things we do. We will bring companies and investments there. and one day we will find an IIt super campus with other research actors and companies to innovate. Every now and then we have to think big. “
Let’s go back to the robot valley and what will be born from this ecosystem. What can we expect?
“I see two ways, one in the short term: a lot of it concerns help in the workplace. To prevent injuries and make efforts instead of the person. Very remote-controlled and not very autonomous, but useful. We can already do this, we have a project with Inail to lift weights. Injury statistics show that 70% relate to the musculoskeletal system, wrong postures and lifting. Here there is a component of human-machine interaction that also concerns artificial intelligence, with tactile sensors to measure the contact and strength. In the long term, on the other hand, we speak of robots sent to the home of an elderly person to assist him. It is a difficult problem that requires a lot of intelligence and dexterity, to move safely in a human environment where there are many unexpected events, this is a problem as complex as the autonomous car. Which in fact works and doesn’t work “.
Musk said “I didn’t think there were so many degrees of freedom” for autonomous driving.
“Exactly, for all the variables that may exist. In both situations you can get hurt, with a car but also with a robot that falls into the house or trips. It’s a heavy AI problem, I’d like to have a project in this direction “.
Here in Rimini iCub is back, remotely controlled by an operator with wearable sensors. What will you make him do in the future?
“Technologies born from iCub are those of remote driving with a three-dimensional viewer and the exoskeleton that measures the position of the fingers and gives tactile feedback. In a factory, a robot can be sent to a dangerous area by guiding it from the desk. those who study how to manipulate objects with a sophisticated hand. Ultimately they don’t necessarily have to look like us, but for us the humanoid form was an interesting challenge. Then there are scientific reasons. In the study of artificial intelligence there are theories that they say that if you want to develop interaction skills with man, a humanoid form is necessary “.
History
The caress of a bionic hand
by Matteo Marini
Are they already applications on the way to industrialization?
“Prosthetic hands, born from the exchange of ideas between those who designed the hand of iCub for example. Some companies were interested but found it complicated, so we simplified it. We have obtained CE certification, we hope that soon a startup will have it. bring to market “.
How many startups were born from Iit?
“We have 28 at the moment. They are born all the time. Last year four were born that raised 16 million in investment from external lenders.”
What do they do?
“Nova vido works on the artificial liquid retina. Photosensitive nanoparticles to be injected on patients suffering from maculopathies. It works on laboratory guinea pigs, the goal is to test them on humans. Foresee biosystems makes sensors for the analysis of drugs, even on ten thousand molecules in once, to understand if they are cardiotoxic, which is the first cause of failure in clinical tests. Screening saves a lot of time and money. The third, Cortical, is fantastic. It makes electrodes for the cerebral cortex, like the ones developed by Elon Musk. But this one has 1024 measuring points and signal amplification. In the first year it already has orders from the research world, there is an incredible hunger for these devices. The latest is a digital pharma combination startup. It is called Iama Therapeutics, it derives from a research in which it was understood that a series of neurodevelopmental pathologies leads to an imbalance in the levels of chlorine in the brain. A drug has been found that r it establishes cognitive function by binding to a protein in the brain but also to a similar one in the kidneys. The doses to the brain are high and it eventually ruins your kidneys. Our people used a number of computational methods to draw an active molecule in the brain and not in the kidneys. Crazy thing. “
He talked about external lenders. How is the IIT financed?
“The state gives us 91.3 million a year. If that seems like a lot, just do a research on the budget of the universities. And then 44 million of external funds, for every euro that comes from the state we found 50 cents. A researcher to have so much extra money he has to keep busy, write projects, prepare them and build them “.
Some time ago he launched the idea of the Alessandro Volta Society. Things?
“We are knocking on various doors to understand if what is happening with the PNRR can become the seed of a sort of Fraunhofer-Gesellschaft, (the German company that brings together dozens of research institutes ndr) which, not by chance, was born with the Marshall Plan. It is a network of institutes where there is an internal exchange with a very practical mission. And where technology transfer is carried out as a main mission, just like the IIT. In the Fraunhofer model, one third of the funding goes to the state and two thirds must come from industry and external planning. We are not very far in my opinion Iit is the demonstration that it can be done. We need to find interlocutors who listen and how to balance the Italian problem of “I don’t get together because I have personal interests”.