"Computers will be good drivers, but they will not remember the past and plan the future while driving"

Says Prof. Idan Segev from the Hebrew University, who will participate today in the event "Discoveries of the future: from the brain to the computer and back" that the National Academy of Sciences will hold today for free in the digital space

"Computers will be good drivers, but they will not remember the past and plan the future while driving." In other words, they will replace us in certain actions but will not be able to take over the world. This is what Prof. Idan Segev from the Hebrew University says in an interview with the Hidan site. who will participate in the event "Discoveries of the future: from the brain to the computer and back" that will be held today by the National Academy of Sciences.

The meeting will be opened and moderated by Prof. David Harel, Vice President of the Academy of Sciences, a leading computer scientist himself. Prof. Segev will speak after him, followed by Prof. Dafna Shaf, also from the Hebrew University, in a lecture entitled "Computers will never be able to do this

Online event - "Discoveries of the future: from the brain to the computer and back", Wednesday, 18/11, 19:00-20:30, on Zoom, on YouTube and on the website of the Israel National Academy of Sciences.

The event will be broadcast on the academy's website, YouTube and Zoom. Participation is free of charge. Zoom viewing requires by pre-registration.

"The brain is a physical example of a computer that can do things. Although a computer is not digital, it is actually a physical system with computational components such as nerve cells. This means that computers can be like us in a way. If we are a certain type of computer that does things such as this phone call. I know people don't like to think of the brain as a calculating machine but we are a special kind of calculating machine after all. "

Even very effective?

Prof. Segev: "Indeed, the brain is an efficient machine in terms of energy. For a 20 watt feed it allows me to talk, sleep, walk. A computer requires a lot of energy. Even my small home computer with a single microchip, unlike the 86 billion "processors" in the brain, requires much more energy. Nature has found a very sophisticated way to build a calculating machine that handles information coming from the world and usually responds correctly, at a very low energy cost. "

"In every period we had other metaphors for the brain. A water pump was once thought to resemble a brain. Then there were holograms, and electrical machines of one sort or another. Today our metaphor is the computer, but this time it is a deep metaphor because both machines perform calculations, and have a deep common denominator. The digital computer is supposed to do every calculation operation and therefore it has to imitate the brain."

"for example In the Blue Brain Project (led by IBM AB) where I am involved, we use the computer to understand brain phenomena. The brain receives information, and performs actions on it. It's a metaphor that becomes more and more similar to us until we think about artificial intelligence, machines that will be similar to us, that's what Dafna will talk about."

The learning computer revolution

The modern computer and especially the new generation of computers which are learning computers are not static but change as a result of the fact that they process information and improve. The whole world of machine learning. Both classical computers and learning computers were inspired by things we learned about the brain. The old digital computers consist of units of zero and one. This is an approximation, albeit not quite accurate, of the operation of the brain, but in the XNUMXs, when the computer was invented, the basic aspect was that it had a logic gate that allowed inputs of zero or one, also induced by the brain, as could be done with the technology of those days .

Until recently, computers were stuck in a static state. Today they include "artificial neurons" and between them there are connections called synapses, named after their counterparts in the brain. Synapse is the changing element in learning processes. Since we learned what happens in the synapse in the brain when it learns, we apply these principles to the digital computer and turn it into a learning machine.

It can be said that our brain gives great inspiration for building computers that are supposed to run the world. To receive information such as autonomous cars that are able to see traffic lights, crosswalks, etc. Today's computers can learn from examples and know how to generalize and create a concept called a crosswalk - any crosswalk, not only one I've seen in the past but also a crosswalk I'll see in the future. The idea of ​​the learning machines also comes from brain inspiration. The computers of the future will start as identical twins but will change thanks to the addition of the computer learning component.

I guess people will ask questions about artificial intelligence, will it replace us? Will computers have awareness, will they be similar enough to us, will they take over us? According to Prof. Segev, the answer to the first question is positive - they will replace us with positions they can perform better. For example driving. The answer to the second question is negative. They will not be able to take over the world."

"We are far from building a machine that will have the emotional capabilities of the brain"

"We are very far from building a machine that will have the abilities of our brain, including emotionality, the ability to contain many things together such as sight and hearing, feelings of speech and movement? For example in the field of movement, today's robots are much more primitive than even a mouse that can jump from place to place. We are very, very far from developing a complete machine in the human sense.

However, in certain areas they will surpass us, such as driving. They will be better drivers than us. it's clear. But at the same time, they won't be able to talk, experience emotions, remember the past and plan the future and all the other things we do when we drive. They will not replace us. They will be good partners in certain places and will allow us to have fewer traffic accidents for example. One day, but it is still far away, they will be close to us."

Many ask if the computers will replace us, what will you answer them?

"The question is in what sense will they replace us? Is it in the broad sense? Because machines have been replacing us for a long time, the bicycle and the car are replacing walking. We used to think that there were things that computers couldn't do - for example being creative - today we can think about utilizing the ability to learn to enable it. But in this sense the revolution of learning machines is on a scale as big or even bigger than the industrial coercion. All three million truck drivers in the US who voted for Trump did so because they are afraid of the future world. If they don't have a job, where will they get a salary?

"There are also many ethical questions surrounding the autonomous cars. If an accident does occur, who is to blame? Who will pay the compensation in case of an accident? The driver, the engineer, the planner? Maybe there won't be a doctor? He will be replaced by a digital doctor who will diagnose the patients. There are lots of new human-machine interactions that need to be handled ethically. And if you have free time, who will pay taxes? There are entire books predicting taxation on robots - so that the state can provide the services? Without a doubt, the learning machines are causing a huge revolution."

"The machines will be individual beings with their own desires. You should not be afraid of them, but you should recognize that it is coming. It is not avoided. You can't elect Trump to prevent them from coming."

the blue brain  

"The personal field I am researching in is the inspiration of technology by the brain, or understanding the brain better through technology. This is the loop that the conference organizers want to make from the computer to the brain and back to the computer. The computer helps us understand ourselves, but we are the ones who develop it."

"Since the brain is such a complex system with many parameters that collide, create an interaction and in the end create a product such as an emotion, feeling or movement, we still do not understand how the complicated machine called the brain works. "

"How to understand my ability to speak now, or how to understand a disease such as Parkinson's. Suddenly a disease happens and something goes wrong in this vast network. These are many small things that go wrong, a small thing here, a small thing there, suddenly, boom, an illness breaks out. It's so multi-parametric that we don't understand."

Epileptic waves

"One of the new approaches, which is controversial but I support it and am part of it, is to try to build a one-to-one exact replica of the brain on the digital computer, this is a great help from the supercomputer that can mathematically represent the processes that take place in the brain - the Blue Brain Project. We can already show preliminary results, for example very detailed imaging - cell by cell and synapse by synapse and get behaviors of the large network that behaves like the neural network in the brain in the case of a disease, not in the sense of consciousness or emotions, but as mentioned to understand the outbreak of a disease such as epilepsy or Parkinson's. We ask when this modeled neural network will behave like a diseased neural network. Because we built this model and entered all the parameters, we see which set of parameters leads to an outbreak of epilepsy, for example, when epileptic waves begin to pass through the network."

control every parameter

"In the Blue Brain project we intend to simulate an entire brain, at this stage of a mouse, we are using a large supercomputer in Switzerland containing over 100 thousand processors and they are working hard to model the cerebral cortex of the mouse, which includes 10 million cells. There too we see outbreaks of disease, or abnormal behavior of the neural network and there we can control each and every parameter and understand which combination that is out of balance causes this disease and which other combination will not cause this disease. In the hope that it will also help us to study the brain, to see if the predictions are correct, but also to develop specific drugs against those parameters, those synapses, those nerve cells that are out of balance.

The supercomputer also allows us to understand our own brain in the sense of computerized imitation or copy." Prof. Segev summarizes.

Here's why, understanding a brain that consumes 20 watts per hour requires resources that even the most powerful supercomputers today struggle to achieve with a large energy consumption of thousands of meters. But eventually it will be inevitable.

More of the topic in Hayadan:

• An international research team has revealed for the first time imaging of neurally active brain tissue
• Israeli scientists participate in the "flagship" project of the European Union for the study of the human brain
• The brain scanner that can read dreams