Carlo van de Weijer ©Fotografie Vincent van den Hoogen
Author profile picture


Innovation Origins regularly speaks to innovation leaders, trendsetters who are high on the innovation ladder. In this series they offer look behind the scenes in their world. Today Carlo van de Weijer has his say. The director of the Eindhoven Artificial Intelligence Systems Institute (EAISI), part of Eindhoven University of Technology (TU/e), is fully immersed in the world of data and robots. From care robots to drones and from football robots to self-propelled cars. “Robots are very good at one particular task, often much better than humans. But will they ever really be able to function like human beings?”

What is EAISI’s strength?

EAISI researches data and artificial intelligence in a variety of ways. Professors from all over the world come to Eindhoven University to work. Students are also involved in EAISI in several ways. “Research on data must eventually be incorporated into the real world”, says Van de Weijer. “The real world is about people. So we need to integrate that intelligence in such a way that people can really make use of it”.

Artificial intelligence is being studied all over the world. “Europe presents itself unlike other regions of the world in this respect,” he continues. For us, people are key. In America, companies are the bosses in this area. I don’t think that will hold up in the long term. At some point, people will rise up against that system. In China, the government is the boss. The power of the people is underestimated there as well, as you now see in Hong Kong. Once people have smelt freedom, they don’t want to go back to the old way of doing things. These are all issues where AI plays a huge role.”

Seeking the added value of the human being

“If you have a job that involves applying rules according to a strict procedure, a computer can probably do a better job. A computer doesn’t make any exceptions. We humans add value when we are willing to stray from those procedures. I do hope that we stick with keeping any deviation from the rules in the hands of human beings. People who skate around the rules can be dangerous, of course. Not everyone obeys the law and that can cause problems at times. But when robots deviate from rules, that’s when things can get really dangerous. Data then becomes a weapon,” Van de Weijer states. This application of data and robots in the real world is an element that professors and students at EAISI need to think about very carefully.”

Is it safer when people bend rules than when computers do it?

“If you can teach a computer that it’s okay to deviate from the rules, where will it end ? We humans care able to bend the rules in a reasonably responsible way. That’s actually attributable to our overall culture of tolerance. If you walk past a red traffic light late at night, there’s nothing to worry about. You’re being silly if you stay put. Sometimes it can even be dangerous if you always follow the rules to the letter. The other day I drove my car in the autonomous driving mode. A car had broken down and had pulled over halfway on the roadside and halfway on the road. The cars in front of me drove across the double yellow line in the middle of the road so as to dodge the car. My car ‘knows’ that it is not allowed to cross that non-passing lane line. Since the brakes wouldn’t work anymore, my car was in danger of crashing into that broken-down car. So in situations like this it’s better to break the rules.”

What could be general rules for robots?

According to Van de Weijer, the rules that sci fi fiction writer Isaac Asimov drew up at the beginning of the last century are still an important guiding principle. “I think we have to define even more of these. God gave Moses ten commandments, which are very similar as well. Perhaps people should become a kind of God for robots,” says Van de Weijer. Even though some of Asimov’s rules are already under pressure. “Drones are flying around in Afghanistan killing people, for the time being remotely controlled by humans,” he explains. “Some people say it’s going to be a long time before robots like this are deployed on a large scale. That may be true. But someday we’re going to have to deal with it, be it another 10 or 100 years. What’s more, robots are already having a huge impact even if they can’t do everything that people do. Look, for instance, at the influence that algorithms are already having on voting behaviour.”

What is humanity’s greatest problem right now?

“That’s a bit further away from robots and AI, but it has to do with the way things are developing. I think we’re suffering from our reptilian brain. We’re too focused on the short term. Occasionally we do invest in the longer term as a natural course of action. As an example, for our offspring. But the time frame that we now should be thinking about, well, we’re not so good at that.”

Which time frame should we be thinking about?

“Are we supposed to want to leave the world behind in good shape for our children? You could also think that they should figure it out themselves. Initiatives like consuming less will never work, I doubt that very much. Unless you clearly benefit from that yourself. Like, for instance, eating less sugar and fat. We are biologically programmed to eat a lot of sugar and fat, because the food supply was scarce and the winters were cold. There’s always plenty of food now, so we don’t need those reserves anymore. Fortunately, our tendency to still really enjoy that kind of food is something that most of us are able to control. Therefore, perhaps we’re wise enough to choose for the longer term when it comes to other issues after all. Even though it sometimes goes against our nature.”

Being right is more important than the truth

“Another consequence of the reptilian brain is that people want to prove that they are right and they’re doing this more and more often. Suppose someone thinks the earth is flat. You used to go to a scientist and they would show you models that proved that the earth is round. Now you have a thousand reports on the internet where five of them say that the earth is flat. So people who believe that the earth is flat cling to those five reports. Also because an algorithm knows that there is a good chance that people will read it if they get to see this report for themselves. That way, they’ll stay glued to the site for a while longer. Somebody like that is constantly being backed up on their own beliefs via the internet. That’s a highly polarizing implication from acting on our instincts.”

“Aside from that, people want to have an opinion just to be able to jump on a bandwagon, no matter what. It gives us a common goal and then we don’t have to look at what’s actually going on. That’s how religion originated, but in essence paper money and the cult of football supporters too. If Feyenoord were to recruit all the Ajax players one by one and vice versa, then the Feyenoord supporters would eventually be cheering just as loudly for those eleven Ajax players. So we originally thought that there’s a difference between one squad and the other squad because they wear different shirts. This phenomenon helps to get large groups of people to move towards the same goal. That’s one of the greatest strengths of humanity, but at the same time it’s extremely dangerous.”

Can computers mimic this behavior?

“Sure they can. If they all receive the same instructions, you can mobilize all that power by setting a specific goal. At some stage, the power of computers is greater than the power of humankind. Except computers are far from capable of doing exactly what we are capable of doing. That will take another 100 or maybe 200 years.”

“Computers are better at chess. But if you put that chess computer next to a coffee machine, it won’t be able to make a cup of coffee. That’s something completely different. It requires a different type of computer. If computers are really going to act like people, they ought to be able to love as well. Will the computer slowly evolve like the way we have evolved? Could some kind of robot emerge with five fingers? We have to be able to pick things up, hold cutlery, etc. It seems to be easier to do that with five fingers. Will the evolved robot develop skin and a 37 degree blood circulation too? That turned out to be the best thing for us after millions of years of evolution. Will robots look just like us in the future? I don’t think it will come to that because robots will concentrate on specific tasks and don’t have to be all-rounders like us. Still, time will tell.”

Is there an innovation that could motivate people to think in the long term?

Van de Weijer doesn’t yet see a technological product that could manage this. He trusts in the power of humans. “If you let your instinct to eat a lot of sugar and fat go free, you’d eat yourself to death before you were 25. So, we do follow rules. There’s a bit of community spirit in us. You can see that in biology, too. Fouling your own nest isn’t good for your offspring. Which means that a little bit of long-term thinking is part of our evolutionary instinct. That’s something that regularly pops up and I think we will have to rely on it in the future as well.”

What is common denominator for success? There are many technological developments that run parallel to each other. What do their successes have in common?

“If the average human being benefits from something, I consider it a success.” In Van de Weijer’s view, the size of a new development or product is not that important; it’s the impact that matters. “You have major developments that have a minor effect and minor developments that have a major effect. Teflon, for example. This material was invented in order to prevent a rocket from burning up when it reenters the atmosphere after a space flight. But we also have saucepans with this material so your eggs won’t burn. For a lot of people, the last application would be more important.”

“If it benefits humanity just a little bit, then you’ve booked some success. Betterment of society can be a goal in itself. We have done the same thing here in the region around Eindhoven. We have reinvented ourselves over and over again. First of all, via computers with tubes. Then with transistors. When the development process had run its course, we came up with microchips. Once these have been fully perfected, we’ll have to think of something else. At some point, we should invest in that process and go from more and more to better, better, better. That’s the essential discussion, I think.”

This article is part of the ‘Innovation leaders’ series. Other articles can be found here