| At the age of 37, Parisian engineer Frédéric Kaplan divides his time between his laboratory at the Swiss Federal Institute of Technology in Lausanne and his company in the centre of the same city. He has a sole objective in mind: design objects that will share their owners’ lives by combining the world of robotics with artificial intelligence.
By Pierre-Yves Frei, 19 january 2011
Are there any two things less alike than a book and a machine? A stone – perhaps. A book has no mechanics. Its pages are turned by the ends of our fingertips. The text does not budge. Leave it in a library for a hundred years and the dust would not make any difference either. Once printed, a book is. It never becomes. Unless, of course, you scribble down notes everywhere. As Kaplan’s father used to do.
We may be accustomed to discourteous offensives on the behalf of novel technologies that have the knack of sculpting, de novo, our day-to-day life without asking our opinion. But a book machine... Even the least conservative may be somewhat offended. The sermon, however, is beguiling. And the preacher possesses the art of words and demonstration. Frédéric Kaplan is 37 but looks younger. Slender, with his head closely shaven, his stride is long and strong – his apparent youth no doubt the result of regular sport. For Kaplan runs. All day long. From one place to another. From the CRAFT laboratory to the Swiss Federal Institute of Technology (EFPL) where he supervises the work of seven doctoral students – who are also fascinated by robotics and artificial intelligence – to his office in the centre of Lausanne where he set up his fledgling company, OZWE, in 2008.
More Parisian than French – as he himself claims – Kaplan is far-seeing. Very far-seeing – a trait suggested by his premises on the 14th and top floor in the Tour Bel-Air where, surrounded by bay windows, the engineer has a view overlooking Lake Geneva hindered only by a lingering mist over each end of the lake.
The Aibo years
The view would grab all your attention were it not for the odd collection of robots, lined up like trophies, which adorn the engineer’s bookshelves. “This one is almost a collector’s piece, remarks Kaplan, as he shows a sublime blue android sporting a marvelous appearance of neglect. “My colleagues at Sony offered it to me when I left the firm in 2006.”
A rudimentary robot. A toy in truth. Nothing comparable with the robots this ex-PhD student of ParisTech’s elite ENST telecoms institute has been working on these past years. When he arrived at Sony France, Kaplan was part of the Aibo adventure, named after the popular robotic pet dog that the Japanese giant firm saw as the first family leisure robot. Unfortunately, Aibo was far more successful in the media than it was in homes. Sony sold barely 200,000 units before deciding that, perhaps after all, the world was not yet ready for the robotics enterprise, canine or otherwise. “This relative lack of success has no doubt something to do with the price of Aibo,” says Kaplan. “But it is also to do with the cultural distrust towards robots which is so characteristic of the Western world. While robots remain robots for the Japanese, we Westerners regard them as splinters in our conviction of what we deem to be our own humanity. In the Western culture, there is no room for hybrids. You are either human, or you are machine. If you are partly both, then you simply do not belong.”
The French engineer is not only interested in machines. He is also fascinated by human relationships. Meticulously and didactically, he explains his concept of how each era’s machines inspire metaphors that serve to describe the working not only of the human body but also of conscience or even intelligence. He uses the example of genetic inheritance, which is so frequently associated with the genetic code, suggesting thereby that “simple” decoding gives an understanding of life. “This metaphor of code, which is used to illustrate the realm of genetics, suffers from serious drawbacks now that we are well aware that the science of genetics implies procedures that are far larger and more complex than those we were aware of only a decade ago.”
Machines as metaphors
The same kind of observation can be made for artificial intelligence. As a science, it sought for many years to approach and even to imitate human intelligence, but ended only in a succession of achievements that demonstrated the exercise’s limits. “For a long time, we thought that the reincarnation of intelligence was a game of chess,” continues Kaplan. “A machine that could beat a chess master could be qualified as intelligent. However, when Deep Blue beat Kasparov it was concluded that the computer was certainly endowed with a singular calculating capacity but, in all good judgment, it could not be described as intelligent.” Phew. A narrow escape for humanity, which remains the sole guardian of real intelligence.
The engineer ventures still further. The only machine that will ever be able to mimic a human being is... another human being. Kaplan is convinced that human intelligence needs a human body. And not only from the very beginning, that is, from the day of its birth – human intelligence also needs to grow with a human body. The surrounding world has to be discovered, and intelligence has to unfold according to what the body allows or forbids. Naturally, our brain commands our movements but, in turn, our body moulds our intelligence by defining the latter’s relationship with the world.
At the heart of human life, and of all other forms of intelligent life, learning is key to relating with the world. An intention, an attempt, a failure. Yet another attempt – a different one – but with the same intention and, this time, it’s a success. The lesson is learnt. This is what robotics has been trying to instill into machines for years. And it has been an ongoing question for Kaplan ever since his days with Sony, which is the best part of a decade.
For robots, the process of learning is very similar. A robot needs something to learn from. Its circuits are programmed to reach an ideal. Let us imagine that one of its tasks is to “walk”. A robot has no knowledge of its body and even less of what could allow it to move forwards. However, it knows that its ideal, for example, is to find a form of harmony between minimum energy consumption and progression. By trial and error, hence by the act of learning, the robot will end up approaching its ideal. “As far as I’m concerned,” continues Kaplan, “I have tried to infuse robots with an ideal to reach: curiosity. For me it was the best way of giving them the largest opening possible onto their environment. And what’s more, it gives them the capacity to take us humans by surprise. If they are capable of disconcerting us, if they are able to evolve despite us or thanks to us, then we might begin to stop seeing them as disembodied objects, and start living together.”
Once again, Kaplan is less concerned with a robot’s degree of sophistication than with following his own path, that is to say, finding a way of putting an end to a one-way relationship in which humans are the sole actors. Is it not a fascinating thought that these objects could, in turn, have an effect on us? Or at least recall our actions so that each of us would end up by having an exclusive relationship with them? Such exclusiveness entails, in particular, a certain capacity of movement on behalf of the objects. Yet again, body and mind are inseparable.
Gestures Kaplan interacts with computers with no mouse, keyboard or tactile screen but with simple gestures.
A decisive encounter
While still working for Sony, Kaplan attended a seminar at the School of Art and Design in Lausanne (ECAL). The sparkling creativity of the school combined with fruitful encounters culminated in a project for a playroom for Aibo, the pet dog robot, where its learning potential would be maximized. Unfortunately, Sony decided to put an end to the Aibo commercial experience at about the same time. Consequently, the playroom suffered the same fate. Kaplan’s trip to Switzerland, however, proved to be decisive – because it brought him together with the industrial designer Martino d’Esposito.
Kaplan returned to his beloved banks of the Seine, but the two men kept in touch. Aibo’s demise had convinced Kaplan that it was time for him to move on. Until then, he had been head of fundamental research in a commercial enterprise; from then on he was to be in charge of applied research in a temple of public science, the Swiss Federal Institute of Technology in Lausanne, within a team dedicated to the field of interactive furniture and robots.
This is where, with his accomplice d’Esposito, Kaplan developed a novel form of computer. Mounted on a robotic arm, the computer is able by way of its screen to follow a person who has just entered the room, much in the way a friendly face would do. It is also able to record the person’s desires as conveyed by hand gestures. The approach is so innovative that in February 2008 the Museum of Modern Art in New York (MoMA) presented the Franco-Swiss prototype in its “Design and the Elastic Mind” exhibition. This critic’s choice encouraged the two men to market their product, and their company OZWE saw the light of day. When Kaplan is asked whether the Wizard of Oz had come to mind when naming his company, he acknowledges: “It’s a great story that asserts the existence of hybrids. Remember the Tin Woodman who’s looking for a heart.”
Kaplan’s robotic mobile computer, baptized QB1, is living its own modest commercial life. The buyers are, for the great majority, large firms that are already dealing with novel technologies. They see in this keyboard less and mouse less robotic computer a promising solution. It’s a niche market that does not entirely satisfy Kaplan. His ambition remains to penetrate homes – every single home – with his philosophy of interactive machines capable of soaking in the peculiarities of each and every one of us, with an end to creating a shared experience.
Read and share
To do this, Kaplan has reversed the logics of his QB1. Instead of suggesting an expensive up-market product with a resolutely futuristic tendency, which may never be altogether understood, he is now offering very simple and cute machines: book machines. “The term is perhaps a little strong, but that’s the idea. They are digital books that go further than the simple experience of reading a static text. You can select any word and get a spontaneous link to an online encyclopedia. You can also type in comments as you read and make them public if you wish. Likewise, authors can continue to intervene in their work and make it known to their readers.” The engineer knows what he is talking about – his own books are already available on the Apple iPad, together with the other hundred or so books offered by Bookapp, an application developed by his firm OZWE and Bread & Butter, a mixed media design agency based in Lausanne.
“The whole adventure is fascinating,” confesses the French engineer. “You can follow, in real time, any request made for any one of the books in our collection. You can also witness their progress and the formation of readership communities.” The young researcher-cum-entrepreneur turns towards the bay window, seemingly lost in thought as he contemplates the view. He then suddenly starts, as reality and the passing of time hits him. “Oh! I must go! I have a meeting!” Run, Kaplan. Run.