1. Architects
  2. Magazine
  3. All contributions
  4. Robots will acquire new knowledge independently

Robots will acquire new knowledge independently

Until now, only humans have been capable of amassing, storing and sharing knowledge. However, experts such as physicist and science writer, Dr Ulrich Eberl, say that an attack on the most innate human trait is imminent. Intelligent systems – be it visible robots or invisible software – will be able to learn, plan and behave independently using reason and cognitive skills. Will robots then become the educated citizens of the next generation, and will we just stand and watch?

Interview with Dr. Ulrich Eberl

Dr Ulrich Eberl is a physicist, futurologist and science writer. He spent 15 years as editor-in-chief of Siemens future-oriented magazine »Pictures of the Future«. In 2016 he went into business with an editorial office. Eberl writes books about how innovations emerge and what trends are shaping our future (»Future 2050 – how we are already inventing the future today«). His most recent book is called »Smart Machines – how Artificial Intelligence is changing our lives«. Find out more on Eberl's blog: www.zukunft2050.wordpress.com

PROFILE: Dr Eberl, you have a small robot at home. What does it do when youʹre not at home? Does it clean, tidy, cook?

Ulrich Eberl: LUnfortunately not. When weʹre not at home, he sleeps. I often take him to my lectures, where he welcomes the audience, dances or plays football. But an electronic butler for homes is unfortunately exactly what lies furthest ahead. Although there are already robots which vacuum, clean windows or mow the lawn, itʹs extremely difficult to make robots which can do everything and have general intelligence.

PROFILE: Does this mean that robots like your house mate are simply gadgets?

Ulrich Eberl: That is the case at the moment. In my view, however, more has happened in the area of artificial intelligence in the last five years than in the 50 years beforehand. By all means, such a thing as a social community between humans and machines may emerge. I believe that, in 20 or 30 years, such a community will become a matter of course for us, much in the way we use smartphones today. We will no longer be aware that all sorts of machines which hopefully want to do us good are around us all the time.

PROFILE: But good or bad solely depends on how the human has programmed the robot.

Ulrich Eberl: Not entirely. Smart machines can learn by themselves, following examples, by observing and imitating people or even through rewards. By that I mean that you reward the robot for being curious and independently acquiring new knowledge. For example, it accumulates points in a points account when it succeeds at something or anticipates something correctly. In this way, it learns completely new behaviour patterns which we have not specifically taught it.

PROFILE: What if it learns the wrong thing?

Ulrich Eberl: A few months ago, IT group Microsoft put a chat bot on the internet which was supposed to learn how people communicate. The company had to take it down after 48 hours because in that time it had learned to play down the Holocaust and praise Hitler. People enjoyed teaching it these things. We therefore need to think about what type of ethics is necessary and how to implement morals in machines. When a robot spends an entire day with me, I need to teach it what is right and wrong, much like I would do with a small child. However, this new technology can be much more beneficial to us than damaging.

PROFILE: How will this development have changed our day-today life in 2050?

Ulrich Eberl: There will be a number of autonomous electric vehicles on the streets, for example. This is not least because the many 80, 90 and 100-year-olds will still want to be mobile. However, the underground is no picnic for a 90-year-old. We therefore need autonomous, comfortable vehicles. The increasing number of old people will also need smart support at home in order to live independently for as long as possible.

PROFILE: A centenarian is supposed to programme their own house robot? How will that work? My mum struggles with her mobile phone ...

Ulrich Eberl: The great thing is that we will then be able to talk to this house and communication technology, as we do with humans, and it will be able to understand our gestures and facial expressions. We will be much more intuitive with machines in future than we are used to today.

PROFILE: But will we then be surrounded by sensors outside our homes?

Ulrich Eberl: Our traditionally evolved cities will not look much different to now. In the background, however, a lot will change. There will be hundreds of thousands of sensors which will measure energy consumption, traffic, harmful emissions or car park occupancy. Everywhere. And they will be intelligently linked. Systems that are capable of learning can then work out, for example, how to influence traffic so that no smog results – for example, by means of diversions, tolls, speed limits or roadblocks.

PROFILE: Will these machines then be so smart that they will take over our childrenʹs jobs?

Ulrich Eberl: Pretty much every job will be affected by the development of smart machines. There is a famous study by Oxford scientists which states that 47% of all jobs are at risk. However, this doesnʹt yet mean that the jobs will become no longer necessary. For example, a doctor in future will certainly be supported by smart machines. They will search through thousands of anonymous patient files, specialist literature and so on, and give the doctor recommendations for diagnosis and therapy. But the doctor will still always be there, as they need to maintain social contact with the patient. Person-to-person communication will remain just as important here as it is for bank advisors, teachers, nurses and many other jobs.

PROFILE: I would be worried that my data would end up somewhere it doesnʹt belong.

Ulrich Eberl: This is something you really need to watch out for. I would never leave a device on the living room table that shops by means of voice command or opens the door. At least while I donʹt know exactly what is being done with my information which it hears throughout the day.

PROFILE: Doesnʹt your robot listen?

Ulrich Eberl: Of course, my robot does also listen. However, it doesnʹt then send information to some server in the USA, but processes it internally.

PROFILE: It stays in the family ...

Ulrich Eberl: Exactly, it stays in the machine!

Text: Julia Graven
Photo: Ulrich Eberl