By Stefan Schmortte und Florian Severin (Messe München Magazine 01/2018)
One of our interviewees develops computers that can think, the other prefers to rely on his gut feeling in his profession. We spoke to Sandra Hirche, professor of information-oriented control, and Michelin-starred chef Tohru Nakamura about artificial intelligence, the virtues of man and machine–and why robots might also have a bad day.
German sculptor Hermann Rosa worked from 1960 onward in the studio that he built himself. The building is a unique feature of the German architectural landscape and the perfect setting for a discussion about human creativity and artificial intelligence. From left to right: Sandra Hirche, Stefan Schmortte, Florian Severin, and Tohru Nakamura.
MM | Sandra Hirche, be honest. How often have you looked at your phone this morning?
Hirche | I’ve looked at it a lot. At least 20 times. It just shows that no one can manage without these devices anymore.
MM | Tohru Nakamura, which would you be more prepared to give up? Your smartphone or your gourmet thermometer for cooking the perfect roast?
Nakamura | Definitely the thermometer, because as a chef, I have to be able to tell how well a piece of meat is cooked in other ways. But I wouldn’t want to give up my smartphone either. Nowadays, we’re all slaves to our devices. It’s always really noticeable when I deliberately put my phone away while I’m on vacation and enjoy the feeling of relaxation which that brings.
Sandra Hirche, born in 1974, is a renowned researcher in the fields of systems control and robotics. She concentrates on haptic assistance in robotics and on control systems in which man and machine interact. She began her academic career at the Technical University in Berlin. Following stints in Tokyo and Australia, she has been a professor of information-oriented control at the Technical University in Munich since 2013.
MM | The smartphone is a relatively harmless example of the way in which our world has become more high-tech. In Japan, where your father comes from, Tohru, a service robot called Pepper has recently begun working as a waiter in a restaurant.
NAKAMURA | People in Japan are very open to new technologies and to robots. But I don’t think a robot like this can replace the communication, the feel, and the conversation that, alongside the food, are essential features of a meal in a restaurant. In future, we need to ensure we preserve some areas that are not defined by machines.
MM | So that waiters continue to receivetips?
Nakamura | It would definitely only be a case of programming the service robot to accept tips, even though tipping isn’t common practice in Japan. I just don’t know whether smart machines would ever have the finesse to recommend the best wine to accompany a specific dish.
Tohru Nakamura, 34, has been the head chef at Geisels Werneckhof in Munich since 2013. As the son of a Japanese father and a German mother, he fuses Asian and European cuisine and is regarded as one of the leading chefs in Germany. In 2016, his artistry was rewarded with a second Michelin star. He is one of only five chefs in Munich to hold two stars.
MM | The management consultants at McKinsey predict that half of all work will be carried out by robots by 2055. What will be left for people to do, Sandra?
Hirche | It’s clear that automation and artificial intelligence will bring about major changes in the world of work. We are developing these machines to make our lives more comfortable and …
MM | ... in the end, we won’t have anymore work to do?
Hirche | Not all work is enjoyable. We are not creating these machines to make people redundant, but to help with their decisions and work processes. That’s my vision of artificial intelligence: machines as assistance systems.
Work gives our lives meaning.Tohru Nakamura
MM | Robots are already a permanent feature of our factories. Now, they are starting to invade our offices as well, because they can do the jobs of fund managers and attorneys. Is this progress?
Nakamura | The question is, of course, how we will define work in future. It’s not just to do with the tasks that we carry out. Work is all about self-esteem. What do you do for a living? is one of the first questions that we ask someone we are just getting to know. Work gives our lives meaning. We need to think carefully about this so that we don’t end up sitting in an empty room somewhere, twiddling our thumbs, and not knowing what to do. We are currently defining the boundaries of machines’ activities in the future. We must not miss out on the opportunity to lay down these limits.
MM | Deep learning has enabled systems based on artificial intelligence to become even smarter. How do machines learn, Sandra?
Hirche | By training. For example, you show the system millions of images of cats and this makes it possible for the system to identify a cat in a picture. That’s down to an algorithm which runs in the background and is nothing like a human brain. My young son only needs three examples to be able to distinguish between a dog and a cat, but the machine has to be shown millions of images. In principle, it simply reproduces the information it has been given in the past.
Nakamura | That is perhaps the key difference between people and machines. We learn something and despite that, we can decide to take a different course in future. We can choose to try something different and surprise a guest with a new culinary creation, for instance. Another example might be a painter who adopts an entirely different style from the one he used 30 years ago. I don’t know whether machines will ever be able to do that. Whether they’ll be able to say “No,” to deliberately make a decision that conflicts with their experience, or to take a risk that could turn out to be a mistake.
Hirche | That’s precisely where the major challenge lies: whether we will succeed in incorporating creativity into an algorithm in future. As things currently stand, machines are not able to match the achievements of human geniuses.
MM | But in the past, a machine was able to beat Garry Kasparov at chess and, more recently, even the current Go world champion. Go is a strategy game that’s so complicated it makes chess look like an improved version of Ludo.
Hirche | That’s true, but the machine ultimately won only as a result of a huge amount of computing power. Another example is the trailer for the science-fiction movie Morgan last year, which was the first trailer in the history of film to be developed and edited independently by a computer. But in this case, IBM’s Watson computer simply analyzed what had been successful and had gone down well with audiences in the past. That’s creativity at a very low level. Lateral thinking, which allows humans to find inspiration in a wide variety of areas, is still very difficult for machines to replicate.
MM | But machines can work with absolute precision. That must be very important in your profession, Tohru?
Nakamura | Of course, we need clearly defined processes and rules to prepare food at our level. Everything must run smoothly. But in addition to all the technology that we use in top-level cuisine, team spirit is also very important. I find it very difficult to imagine a robot working alongside us and doing the same tasks. How can we create a team consisting of a person and a machine? Or from a more practical perspective, what should we do with Pepper, our service robot, on our team outing? Should we take it with us?
MM | It could at least go and get the car.
Nakamura | That’s a good idea, but would we be able to accept Pepper as a member of our team? I think that the more intelligent these systems become, the more we need to ask ourselves how we intend to integrate them into society. Precision is all very well, but this is where the human factor, the ability to constantly question or be critical of oneself, to have a bad day, or to make a mistake, falls by the wayside.
MM | Are mistakes a positive thing?
Nakamura | Of course, mistakes should never get as far as the guest in the restaurant. But mistakes can happen in top kitchens just as they can anywhere else. And then you get annoyed and do everything possible not to make the same mistake again.
MM | Do machines also get annoyed, Sandra?
Hirche | You can program a robot so that it constantly tries to receive praise. If that doesn’t happen, the robot gets annoyed, not in a human way but in an algorithmic way. It gets a smack on the hand, so to speak, when it makes a mistake.
I don’t believe in a future where people become slaves to machines.Sandra Hirche
MM | Mistakes waste time and cost money. Wouldn’t it be wonderful if in future we could make fewer mistakes in the highly efficient world of business?
Nakamura | But we would also have to spend money on psychiatrists to treat us because our nerves would be utterly ruined [he laughs]. I’m certain that none of us would like to live in a world where there were no mistakes.
MM | But?
Hirche | Perfection and efficiency have a place in our society. But we would not want to give up what is genuinely human, which includes creativity and empathy, but also our mistakes. They are part of our fundamental nature.
MM | Technology fans are predicting that this is precisely why people will become increasingly superfluous and will simply be there to facilitate an evolution that has the ultimate goal of creating an artificial super-intelligence. Are the machines likely to dominate us as they do in many science- fiction movies?
Hirche | I don’t believe in a future where people become slaves to machines. We will always be in a position to pull the plug. The question is also whether we really want a world like that. Smartphones weren’t developed to encourage social isolation, but to improve communication. The situation with artificial intelligence is similar. It has advantages and disadvantages, which is why we need to think very carefully now about the extent to which we want to make use of this technology …
Nakamura | ... and who will be in control of these things. We have made huge technological progress in the last ten or 20 years.
Hirche | That’s very true. I recently tried an augmented reality headset and was really amazed by the quality and by how realistically things can be represented. But we are in a position to decide for ourselves how far we want these developments to go.
It’s amazing how a wink from a machine can give rise to feelings in humans.Sandra Hirche
MM | Perhaps as far as the humanoid robots in Hollywood movies that we will fall in love with?
Hirche | We definitely won’t be falling in love with machines, but experiments have been carried out that show that people can have emotional relationships with machines. For example, a small cuddly seal robot called Paro has been developed in Japan to help autistic children to respond to touch and to make social contacts.
MM | So perhaps the perfect wife or husband isn’t so far away?
Hirche | I can’t imagine that happening, because I would always be aware that it was a programmed machine ...
Nakamura | ... you could have different models: empathetic or sporty, for example.
Hirche | [she laughs] That would be a possibility. But seriously, it’s amazing how a wink from a machine can give rise to feelings in humans. Personally, I think that these robot dolls are not worth working on. But the humanization of technology that we’re now seeing is perhaps part of the very nature of things, because humans prefer to interact with other humans.
One interesting question is how we will define the normal condition of the human body in future.Tohru Nakamura
MM | This seems to be working the other way around, too. Top Paralympic athletes with their high-tech prostheses are jumping higher and further than ever before. Are people and machines merging together?
Hirche | We could see this sort of symbiosis in future and not only at the top levels of sport. An artificial pancreas is already available for diabetics that automatically measures their blood sugar and uses an insulin pump to maintain the right balance. We can use machines to replace functions that we have lost. In future, they may even be able to improve human bodies in the same way.
Nakamura | We have been doing this for centuries with spectacles. One interesting question is how we will define the normal condition of the human body in future. Will it still be OK if one arm is two millimeters longer than the other? Do I have to have 20/20 vision or is it acceptable if my sight isn’t quite as good as that?
Hirche | We will begin by offering technical solutions to compensate for missing functions. It is hard to say today where this is likely to end.
Nakamura | I just hope we’ll have the opportunity to get used to these things gradually. The merging of humans and machines certainly won’t happen overnight. In the same way as with self-driving vehicles, the developments will happen slowly. There will be initial tests, road trials, and then routes that will be approved. We need a step-by-step approach.
MM | In his worldwide bestseller Homo Deus, the Israeli historian Yuval Noah Harari writes that optimizing the human body will be the greatest revolution since the beginning of life on our planet. Is this revolution desirable?
Nakamura | Desires can change and are always a question of how you are personally influenced by a situation. If there is a possibility that you will lose a loved one, you will make a different judgment from someone who is unaffected.
Hirche | If I have health problems in 30 or 40 years’ time, I will be very grateful for help of this kind and I will definitely make use of it. In societal terms, we need to ask ourselves whether we want this optimized superhuman or whether we’d prefer to go on living with our individual differences, including our problems.
Nakamura | I think we run the risk of raising the standard of what is normal even higher because of the technology available to us. That could increasingly lead to people being penalized. For example, if health insurance is made dependent on people’s lifestyles in future, that would have a kind of conditioning effect. If you’re a non-smoker and take regular exercise, you’d pay less.
MM | In China this vision will soon become reality. By 2020, the Chinese state aims to have recorded data about all its citizens, from top to toe, so to speak. People who behave well will benefit from this, while people who misbehave will be investigated. Would you provide your data for a system like this?
Hirche | Only if it helped improve my life. I’m happy to make my location data available to Google, because it allows me to avoid the next traffic jam. If I can benefit in medical terms, I’d be prepared to provide my health records, because this could make it easier to identify symptoms and prescribe the right medication and other treatments. It’s simply a question of what I get out of it in each individual case ...
Nakamura | ... and who will have access to my data in future. The progress that we can make as a result of our technological achievements is a positive development, but only if it is democratic.
MM | What sort of machine with artificial intelligence would you like to have?
Hirche | An all-round domestic assistant, but that’s a long way off.
Nakamura | I’d like a washing system that magically washes my dirty clothes, irons them, and puts them back in the cupboard just where I want them.
Hirche | My all-round domestic machine would, of course, be able to do that, too.
MM | What about cooking?
Hirche | I’m more than happy to leave that to top chefs like Tohru Nakamura.