Robotization, automation and artificial intelligence will soon change the world that we live in. But the worries that robots could take over are not warranted. Michal Pěchouček, the most famous Czech AI expert who currently works for the security company Avast, is convinced of it too. He found time in his packed schedule and in an interview with LP-Life.cz he spoke openly about things such as how automation could make up to two billion people lose their jobs. At the same time, AI will help detect diseases such as cancer or glaucoma earlier. And the endless circle of going from doctor to doctor will end, because a smart program will diagnose your illness in an instant. What's fascinating is that this all can be a reality in a mere few years.
Many people worry that robots will come and they'll control and command us. Thanks to being superintelligent and therefore more capable than humans, they'll have dominion over us. Unlike Elon Musk, I do not worry about this. But I think that there are two big problems where AI can endanger our lives.
One of them is manipulation within the cyberspace, manipulating people using the spread of information on the internet and creating opinions to order. These days, AI can predict our preferences, desires or interests and thanks to that offers us relevant advertising. And if AI starts to be able to identify what kind of tests you like and what you like about them, it will create a text with a hidden message to order. It will also be able to elicit a positive emotion regarding certain information, and thus will be able to manipulate you. These days, international conflicts are waged in cyberspace and AI will play a big role in this.
Thanks to automation, a large number of jobs will disappear as well, people cite up to two billion jobs, so about a half of the entire job market on the planet. And people will lose their jobs like this. It's not the first time there's been technological revolutions that have massively changed the job market, but all of those were gradual and took a longer time, it was possible to prepare for the change. This change is so fast that it's not possible to change the education system to teach our children whatever will be needed on the job market when they're done studying, because we don't know what that will be. And I'm afraid that thanks to the speed of change and our unpreparedness for it, the scissors of income inequality between the rich and the poor will keep opening and society will not be able to respond to this huge societal change.
So you don't see any potential in the scenarios from sci-fi novels and movies, where robots gain sentience and rebel against humans?
I'm a scientist, so I can say that there is neither proof nor definite argument proving that that can never happen. So it could happen, but at the same time it is unlikely and does not deserve that much thought. It's good to focus on scenarios that are more likely.
Can you determine the things Čapek was wrong about in his play R.U.R.? In it robots rebelled against people and were able to fall in love...
The essence of AI that can rebel against humanity does not need to have the form of hardware at all, because that is very tasking, complicated and pointless. If AI wanted to control people, it'd do so through people. It would just manipulate people to do what it wants. So the entire concept of a robot uprising is a romanticized version that serves it to people in an easier-to-imagine form.
AI will not be evolving towards the creation of superintelligent physical robots, so I do not expect more complicated mental states. What seems more likely to me is a human falling in love with some kind of algorithm on social media. That is quite likely.
A few years ago it was said that AI cannot defeat a human in chess, then that got busted and people said it could not win in the Japanese game Go. And then that got busted too. How far can it go?
All complicated games have mostly been solved already. Algorithms are better than a human and that's because of calculations, high performance and also the quality of scientific work. These days, scientists are interested in replicating the mental processes of intelligent people in highly unpredictable environments, such as the thoughts of a doctor, the ability of a safety analyst to analyze the behavior of malware in a network. Scientists are interested in the process of turning specialized machines into something that possesses general artificial intelligence. So, for example creating an algorithm that's able to play both chess and go. These are challenging tasks.
That's all speculation again. Robots will definitely be performing some surgeries, but I don't consider that a revolutionary thing. Robot-led surgeries are a thing already. But that does not influence anything in the way the implementation of machine learning and diagnostics does. These days, 25 percent of US finances goes to doctors' salaries and those do diagnostics most of the time. And those are the kind of jobs that can be automated.
I think that for interesting and complicated operations you always need a human who will always be better. But the moment we're able to gather samples and data history and launch artificial intelligence, the AI can not only diagnose things well, but predict them too. And that's exactly where AI can come into its own. It can give access to healthcare to a large number of people who do not have it at the moment. Not by implementing robots, thought, but by the automation of diagnoses and letting doctors have time to really cure people.
The best cure to cancer is predicting it. If it's possible to predict glaucoma from a high-quality photo of a human retina 10 years in advance, if we were able to do something like that with cancer, we can do a lot to prevent it from developing at all.
Transportation will change and drivers will practically cease to exist, that will be a job of the past. It will be way more effective, easier and safer to dispatch self-driving cars. Today we find it unimaginable, but there the change will be the fastest actually. Technically it's not complicated at all and I think that self-driving cars will start coming into cities in under ten years. In massive numbers within 10 to 20 years. Only implementing them in remote locations will be a long-term process.
A plane does pilot itself, the pilot only takes off and lands. But the air transport industry is dealing with problems other than automation right now, most of all safety and the carbon footprint. People will want to solve those first.
You've mentioned unemployment that will be brought about by robotization and automation. Up to two billion people will lose their jobs and many fear that. How is it going to be solved?
The concept of jobs will change, they will cease being the DNA of our personality. People won't stay in one field for life, they will be forced to retrain. You as a journalist probably already know you'll have to find something else to do, because articles will be written by artificial intelligence.
Jobs will also come to have a broader definition, it won't only be the thing that defines economic value. It will include things such as raising children, caring for aging parents, volunteering at school, the town, even retraining will be a job, partially. We can combat the disappearance of jobs by trying to keep as many jobs for people in society as possible - jobs in social services, healthcare, education, hospitality, tourism, simply anywhere where people would like to be surrounded by people. Nobody is going to want a robot caring for people at a hospital, everyone will want a nurse.
Developed countries that will have enough jobs with added economic value will be able to afford nurses in hospitals. Countries that will miss out on this change will be poor, economically dependent and will be unable to afford paying nurses and will have robots working in hospitals.
These days it's developers in the field of artificial intelligence, data analysts, people who construct robots and generally work in really hi-tech fields. On the other side we have lawyers, finance analysts, clerks and other similar professions that are going to be automated.
It's all fields with monotonous, manual labor that can be automated. Rather than the robots replacing people in factories, I'm afraid that there will be a decline in office jobs - clerks, managers, accountants, insurance agents, bank clerks. It can happen to everyone who goes to work and encounters artificially inflated employment in their sector.
How will Czechia fare? Do we have what it takes to be a rich and technologically developed country so that we do not have to have robots in hospitals?
Czechia does have what it takes. It has smart people, technology experts and researchers are coming back to Prague after gaining experience overseas. We're one of the few countries in Europe that has a government strategy regarding artificial intelligence and the Czech government is among those that take artificial intelligence seriously. On the other hand they underestimate the potential disappearance of jobs. They lack a vision.
As inhabitants of Prague we don't quite realize that it already is that city in many aspects. Prague is simply a magnet, there's interesting stuff here. And it's attractive to Czech people returning from abroad. Even Slovaks who have returned to Bratislava usually get "stuck" in Prague. It's making Prague an incredible melting pot of talent and it should be able to make use of that. Us academics already know that these "returned" scientists create incredible value here, because they have overseas connections. That's why we started the movement prg.ai, which connects Charles University, Czech Technical University and The Institute of Computer Science of the Czech Academy of Sciences which aims to make Prague an attractive destination for AI talents. So that people, who want to move to Canada or California in pursuit of AI, would consider Prague too.
That'd be fantastic, I'd love that. But there's a long road ahead until then. For the start I want us to have some AI business here and to allow people who make AI to work here. Once we have many AI companies, experts and scientists, it will be possible to create ideas that will move the world.
We're small. We have to be bigger, have more of everything. More AI, more tax breaks, more smart immigrants and faster granting of visas to those that want to create AI. In Canada, when you're an AI scientist, you'll receive a visa in under seven days. Personally I was trying to get an AI scientist to ČVUT, it took me a year and it didn't work out. This way we won't become an AI power.
In your fast confession you mentioned in a few years, a robot could be indistinguishable from a human. Is that possible?
I don't think it's anywhere close. On top of that it's not useful, that's why it will take long to become a reality. It's the same as sending a human to Mars - it's nonsense, expensive, risky and realistically won't get us anywhere. It's this ego thing. It's better to send a robot that will collect data, samples and explore.
What will be the required changes in legislature? What if a car or a robot accidentally kills a human, who is supposed to be responsible?
There's ethical research on robotic responsibility, but I don't find it that fascinating. There already is legislature blaming somebody for a malfunctioning elevator, if it chops off my hand, somebody is responsible for that. And they'll be held responsible in the same way if a robotic car makes a mistake.
What's interesting is legislature regarding the properties of algorithms that influence our lives. I believe that in the future lawmakers will want to legislatively limit the kinds of AI that can affect people. For example, algorithms that drive cars must have different properties from those that make decisions about medication or invest our money. There's also the interesting questions of the degree to which algorithms can limit a human's free will. Right now it shows that the degree of autonomy of algorithms on the internet "forces" you to buy things that you didn't originally want, and that as a result limits your free will.