“In a few years we will have technology capable of ‘cloning’ human motor intelligence and adapting it to everything.”
The scientist who creates humanoid robots capable of moving in an environment, manipulating objects, playing football. AND director of Google DeepMind’s robotics labone of the most cutting-edge companies on the planet, controlled by Alphabet. Apply artificial intelligence to robotics. And it aims to create robots with human-like intelligence, capable of learning intuitively and doing anything. The field in which we move is that of Artificial General Intelligence (AGI). Generality is the machine’s ability to adapt and understand how to perform a task in different contexts.
He is Francesco Nori, 48 years old, from Paduabehind him an extraordinary careerItalian Institute of Technology in Genoa where he created iCub, the first humanoid robot made in Italy.
Then came the call – irresistible – from Google DeepMind to London. Today he is among the 5 directors who coordinate all of the company’s robotics, born from the merger of Google Brain (the AI part of Google), and DeepMind, an artificial intelligence startup founded by Demis Hassabis and acquired by Google in 2014.
“We’re trying to communicate with robots in a simple way, exactly like you do with ChatGPT or Gemini. If I tell the language model: create a sentence with all the words that start with the letter B, it does it. In practice I use the way natural communication between two people. I give the robot the ability to understand what it has to do in an intuitive way, offering it a text, an image or an example so that it learns what it has to do“.
Mathematical models are no longer used but you have a data driven approach, which allows the robot to self-develop its skills and understand how to move. “We start with very few assumptions and try to take data from the environment or generate them with simulation systems. We tell the robot: you have to win the game and it learns everything on its own. It learns to coordinate its movements, to dribble , to defend the goal and score goals. Moving from mathematical models to data means greatly improving the ability to adapt.”
Beautiful Minds
Luca Berdondini: “My chip to be implanted in the brain is more powerful than Musk’s”
by Eleonora Chioda
Why teach a robot to play football?
“It was an academic exercise, we had enormous media coverage, people loved it. We are now on the cover of Science Robotics, we ended up on an American TV show. It’s a test to understand what a robot can do in everyday life. We are focusing on manipulation, the ability to take objects and put them in a box, cleaning a room, helping an elderly person get out of a chair, all the way to industrial assemblies.”
Beautiful Minds
Damiano De Felice, the Italian who fights antibiotic resistance
by Eleonora Chioda
Nori’s story starts from Padua. He graduated in what was then called computer engineering, today it is computer science. He is doing a doctorate in a theoretical field, that of automation, but for his thesis he deals with a topic between robotics and neuroscience. “I had done a study to understand how the motor control system is organized in some mammals. My thesis was a sort of vocabulary of ‘motor primitives’, that is, essential movements which, when combined together, could create more complex things.
Once the thesis is finished, Nori knows Giulio Sandini, professor of bioengineering and founding director of the Italian Institute of Technology (IIT) e Giorgio Metta.
“We were among the first to be hired. Giulio Sandini was my first mentor, Giorgio Metta, now director of the IIT, is a very dear friend of mine. We were immediately inspired by children. We started studying how they develop certain motor skills and created the first humanoid robot iCub. We taught him to touch things, to lean, to use the whole body to make movements in the environment, to learn Tai Chi.”
Nori stayed at IIT for 14 years, followed many projects, had a great career, and fell in love with the sea. “Genoa stole my heart”. Then unexpected comes the call from Google DeepMind.
“I was at the airport returning from a conference. A friend calls me: “Google DeepMind is looking for a person to start the robotics area.” I had no intention of leaving IIT, but I started having a series of interviews with the CEO, Demis Hassabis and with other important roles. More than getting hired, I was interested in a scientific conversation the offer exceeded all my expectations. They pampered me, listened to me. Then salary, benefits and the chance to work in such a fascinating field.”
Beautiful Minds
Antonio Tomarchio, the prophet of AI: “Many lights and a shadow, it is in the hands of a few”
by Eleonora Chioda
But can you make a robot do anything? “Yes. The possibility of having robots that do everything that humans do is imminent.” Then Nori stops and adds: “I am not so naïve as to think that the technology I am developing will only be used for the good of the world. All technologies must be scary. Think about fire: it is a simple technology, which can cause damage, it can kill people, but we have understood how to use it for good. Think about the car: it is very dangerous, yet we are all happy to use it. We gave ourselves rules, codes, social conventions, traffic lights, stripes, and it became part of our life”
The theme of AI is also the speed at which it grows. “Its computing capacity has exploded like no other technology. The order of magnitude is in the trillion scale. The technology has improved a trillion times over the last 20 years. Where will we get to? We have arrived at a computing capacity that resembles , by order of magnitude, to a biological neural network.”
Robots that will be able to invent anything, even probably what has not yet been invented. “We’ve seen this with language models: we can create completely new, improbable images that don’t exist. There will always be something that he has never been shown that he will do…”
Beautiful Minds
Lucilla Sioli, the creator of AI ACT: “Long work on a black box”
by Eleonora Chioda
Nori has learned many lessons along the way, but the main one concernshumility. “When we started studying artificial intelligence, we thought that language was the result of an interaction with the environment, that is, of understanding and agreeing with the interlocutor what is being talked about. In reality, language is still like this for man, but we have proven that word statistics are sufficient to create language models. Infinite examples of words are enough to have a machine that produces language in a meaningful way. This is a very fascinating exercise because it brings us back down to earth, it makes us understand more about ourselves. And it teaches us to be more humble…”
New father, Nori does not fear for his son’s personal future but rather for the world in which he will live. “Wars, continuous hatred between various nations, global warming. I’m not worried about technology, but about human nature.”
On why technology is being pushed so far, he cites the importance of the impact on society. “Google DeepMind started by putting artificial intelligence into video games and demonstrating that the technology could solve complex problems, without knowing the rules of the game and learning new ways to win. From gaming to life. Those same technologies were then applied to predict the retinopathies or to develop AlphaFold, an artificial intelligence program that helps predict the three-dimensional structure of proteins, which represents one of the great challenges of biology. Predicting the three-dimensional structure of proteins allows us a greater understanding of human health, disease and disease. environment. To do this, in the past it was necessary to carry out analyzes that required 3-4 years of work. Today, thanks to artificial intelligence, everything is done in one day.”
However, if you ask Nori about his motivation, he replies that he is experiencing an internal conflict. “A few years ago I was in New Zealand on a dinghy in the middle of nature and a person very far from my world said to me: ‘Why are you doing this, why do you want to progress to this point?'”
I do not know. I live between these two poles. On the one hand I have a profound interest in technology and knowledge, on the other I ask myself: where will a hyper-technological future take us? Should we just stop and say are we happy like this? We have numerous new problems, global warming, understanding how to use technology for a better future, avoiding becoming an invasive species for the planet…”.
Beautiful Minds
Federica Agosta, the neurologist who trains algorithms to defeat Alzheimer’s
by Eleonora Chioda
He recommends seeing a documentary for young innovators and those who develop algorithms and technology. “It’s called Plug & Pray: it tells the story of an MIT researcher who at the end of his scientific career looks back and sees that many of the things he created have been used for military applications. It’s a documentary that makes you think. Who is dedicated to technological development should aspire to a positive use of technology to help people live better and not to create even more hatred or even more death, if we as a society had always acted according to these dictates, perhaps we would not have created such dangerous weapons … “.
And to all of us he says: “Let’s make society regulate the use of technologiesand let each of us contribute to spreading its ethical use.”
Beautiful Minds
Maurizio Vrenna, eco designer in China “Come and see the unbridled consumerism and you will learn to produce”
by Eleonora Chioda
#Francesco #Nori #Google #DeepMInd #scientist #creates #robots #capable #learning
– 2024-04-23 05:53:49
