Is the “Robolution” dangerous? Robot can’t make coffee
Recently two events have retained my attention.
Alphago, the super computer of Google beat the best GO-player in the world, 4 rounds to 1. And Tay, Microspft’s artificial intelligence,has gone in a day of man’s best friend, designed to interact and learn with him, in a racist digital parrot, fascist, anti-Semitic, pro-Nazi, anti-feminist , pro-kukluxklan. We wondered if artificial intelligence could become president of the USA, we see it already at a candidate … And now return the fear of technological horror
I invite you to a trip in the near future. Personally, I plunge with delight, without minimizing the risks and pitfalls. I’ll talk about employment, artificial intelligence, robots, technology, but in reality I’m going to talk about us …
1 Employment and robots
Media regularly titrate with horror and great detail on how robots will bring unemployment, poverty and the end of humanity. These studies exploit the seam of the uprising of machines, Skynet (the powerful IA that controls humanity in Terminator), or the Matrix, to the scene of the arrival of robots and artificial intelligence. From Oxford University or the Davos’ World Economic Forum, through the Bank of England, studies explain from 40 to 60% of jobs will be occupied by robots in the next 1à years.
Do not throw away, the cut is more than full. The robots will replace us, at work, in relationships and even procreation, through spermbots, nanorobots that will help fertilization …
Forgotten Schumpeter, Solow, Sauvy …
OK … and if we looked at things differently?
Let’s look at evolution. A man can carry on average 30 kg. With the horse, it goes to 1 ton, with the cart, 2 tons. Train: 100 tonnes. The first trucks: 500 tonnes. And the last gear of more or less automated sites may be up to 20 000 tonnes. In 30,000 years it has risen from 30kg to 20000 tonnes.
And during that time, the human population has increased dramatically, civilizations were born, others died, cities, nations, empires are built. And there was … work.
We look these potential technological advances with a bad eye. Have we all become new Luddites? Should we not see positively the potential release of humans 40, 50, 70 hours of repetitive tasks every week? Is it not something to promote rather than fear?
The technology has disrupted almost all sectors and industries, causing, sometimes against their will, into the 21st century. Inevitably, the work will face disruptions. Low-skilled jobs are at risk, as well as some more professional jobs (if we will still need a lawyer to plead, preparing his argument may actually be assisted by an AI rather than white collar …) . But many of the jobs that will disappear are those who simply are more relevant to today’s society. How many there yet grinders? Or iceman for food preservation? And the average age of workers in American industry is 56 years. Young people do not want this type of job. Moreover, a McKinsey study of 2014 shows that it is already possible to replace 20% of the work of a CEO and 80% of that of a cashier. If everything was automated today, 45% of the population would be left unemployed. But we are far.
For if the gradual replacement of humans by machines to do (better and faster) more or less simple tasks, grunt, difficult, …, is inevitable, this is so since the beginning of history. It is even the main reason for the innovation efforts and alleviate the increasing human capacity, so eventually replace human labor, or at least easier. Agricultural machinery has increased tenfold the power of working peasants, and production increased along with the decrease of the peasant population; the development of the automobile has destroyed the ecosystem of the horse, while developing the hand of working out, itself gradually replaced by robots on assembly lines; brokers in the trash instead of the Exchange turned into computer algorithms, without preventing the development of the financialization of the economy …
The machines in the broadest sense, and have destroyed much of the job. And in all sectors.
But then how to explain that the overall number of jobs has increased? How to analyze the significantly lower rate of unemployment in Germany, Korea and Japan, largely automated country than France, under robotic equipment?
Many studies have explored the relationship between employment and unemployment, and showed that in the long run, there is no correspondence between innovation and unemployment. Sauvy (La Machine and Unemployment) dismantled the illusion of “technological unemployment” if in the short term, the machine reduces employment in one sector (eg agriculture), it also has positive effects, lower costs resulting in a transfer to other products (concept of “tipping effect”).
To support these claims, I refer you to the study of Deloitte Cabinet of summer 2015, analyzing 140 years of technological evolution in England and Wales, finds a positive impact on job creation: 1871 to 2011, there is a drastic reduction in activity bleachers with the arrival of washing machines, the decline in the agricultural labor force, but for example, an extraordinary increase in hairdressers, 1 barber with 1793 inhabitants in 1871 1 287 in 2011, or service occupations around the “care” in general (by health professionals, educational, home support …) which rose from 1.1% of the active population to 23.7% .
Yes the age of the machine is there. For nearly 30 years, many trades have died, others are processed, and others created. At the same time, the French population increased by 13.7 million; active population of 6 million, and the unemployment 3 million. Yes jobs were not created. But what concerns us, is that because of automation? Or, by comparing with Japan, Korea or Germany, would it not conversely, because of a lack of automation: not enough productivity, stagnation or increase in labor costs , fewer transfers to other sectors. A vicious circle instead of a virtuous circle.
So announce a number of jobs destroyed by ongoing technological innovations, isolating macroeconomic relations, might have only one objective: to buzz.
2 In Artificial Intelligence, what is important is artificial.
Alpha Go to Google DeepMind tackled the most complex game, GO. Until then, the machines beat the best human chess, checkers, Othello and Jeopardy! But no machine had beaten the best that significantly more complex game than chess. DeepMind broke the last barrier of the smart play. The machine terrace man, 2 games to 0, 9 innings to 1. An event that shook the world of artificial intelligence, as strongly as the Deep Blue IBM battle against Gary Kasparov in Chess. And our favorite cassandre are still plenty to be happy. Forgetting that last fight … date of 1997, it will have taken nearly 20 years to get to GO … It is therefore not so fast.
Now then, if we listen to the comments, humanity is ruined. Machine 9 — human 1. Damned, I tell you. OK. Let’s step aside. What can we see?
Deep Mind learned a lot for the game, and between the October match against Fen Hui, the best European, and the latter competition. The computer was “confident” in his chances of victory in the mid-games played, thanks to probability calculations. But behind DeepMind there are … men. Talented engineers. And also a Go player … Indeed, Fan Hui, after his lost match, joined the team DeepMind. So it’s not an artificial intelligence that wins, but the sum between computing power, human ingenuity and professionalism of a top player.
DeepMind has played shots are interesting and disturbing. “This is not a human movements. I’ve never seen a man play this kind of movement”, said Fan Hui. The computer has an increasingly playing style “creative” than other gaming software of this type. Or more than humans. But is this a real creativity when you know you made him memorize nearly 30 million movements and played against him-even for months?
So we may be uncomfortable with the idea of the supremacy of AI. But besides the fact that the AI does not know “that” play GO, there is a point that no statement or almost, which is the whole point of this experience and all the contributions that the AI can have on humanity.
Fan Hui, triple European champion and Go first top player defeated by DeepMind has served as an advisor to the machine’s team. In the match with Lee Sedol, he did not know how to analyze some amazing shots. But after about ten seconds, he said, he made the relations between the different movements that DeepMind had already played. Fan Hui has the advantage of watching the machine for months to play again and again. And as he played every game against him, during those five months, he saw the machine better. But it has also improved himself, what changed his view of the game. In October, he was ranked 600 in the world e. Now it gravitates in the 300. And if Fan Hui has enabled DeepMind to improve, he learned, too, to be a better player.
Okay these AI may be smart but without us … nothing. Even those self-learning. They must be designed to. And who makes that designs? Deep Learning is powerful, but still not enough for example to play against multiple players. An article in Slate says that from three players, “the concept of ‘good shot’ for every situation is shelved and the delicate strategic balance is upset. The strategies of the three players become interdependent and now rely much on statistical considerations and psychological”. Situation we found at Bridge: it’s a team game, or the computer can not play collective and within it there is a good chance that the machine loses because humans will band together against her. Well, well …
And the great game loser, is perhaps not the Man. But Korea, which has lost ground in the race for the AI and which will invest 765 million Euros in research …
Let’s step aside: the winner is perhaps not Google or Alpha GO. The first winner is Demis Habassis, the boss of Deep Mind, acquired by Google. The second winner is Fan Hui. A designer, a player. 2 men. Extrapolate. Watson, IBM AI is better diagnostician than doctors. Google Car has fewer accidents than normal cars. Safari or Google search know better than us. OK, but what interest if it is not for us? If the AI does not allow us to become better?
3 The robot is it the future of man
Robots can do everything, except to replace us. Today robotics, automation, what is it? Articulated arms, sensors, calculators, electronic objects assigned to a task or set of tasks (paint, screw, move, count, count, wear …). Chips and algorithms in plastic housings which calculate faster and better than us. Optical drives of bar codes, licensing, postal code reading faster and better than us. AI can win against best players of Go, chess, checkers, scrabble, rummy, Black Jack, 421, small horses (but none knows how to play hopscotch, and I expect to see a football game between robots and Barça or a neighborhood team). Super computers are capable of creating a stock market crash in a fraction of a millisecond when man starts a white Monday to finish with a black Friday. So yes, today with robots, co-bots, automation is going to be accelerating. But, it’s not so simple.
Beyond the employment impact of the technologies that still tends to hurt and analyze these studies promise us a future so dark we should stop everything and go back to the ox carts, it seems not seen the huge beam, only stopping as many straws (very numerous indeed). The robot can do a job better than man, excellently. But a task. Or a coherent whole.
Ask Deep Blue beat chess man and he will. But ask him to beat the checkers … No it is programmed for chess, despite the self-learning AI is still human programmation. Ask Watson win Jeopardy. He will. But it’s still the man who programmed. But ask him to play chess … Ask the AlphaGo to play Go, but poker? … Well, ask GoogleCar for driving on snow or fog …
So straws are very numerous but this beam, frankly, you still do not see?
Come on, I help you. Do you know Roomba? A robotic vacuum cleaner, the best selling robot in the world to clean your floor. Overall it is not bad. But it will remain some dust here and there. Ask Atlas to move the brush, it’s funny at first so he left and frankly embarrassing. And if you also ask him to make coffee, put the roast in the oven and do the dishes without breaking anything, so … But it was made for the military, so he had to be ultra efficient …
Automation knows supplement our arms. She knows compensate a part of our brain. Articles are already written by robots. Medical advice can be robotic. The oral argument preparation work can be done with automated systems. The purchase course, part of the pipe, manufacturing of sensitive elements, writing a book … But try asking a robot to make you a coffee. Moreover, China, 2 restaurants on the 3 servers using only robots have closed. And the third, to get by, dismissed his robots …
Robots can distribute drugs. They can clean a room. But not the best prepare for the patient comes out of the operating room. Personal service can. Nurses also. The robots are capable of making a medical diagnosis and dosage. But if there are complications and wrong dosage, the robot can he react? A doctor, yes. Internet, Google and Wikipedia provide access to a tremendous amount of knowledge. But recoilless, without explanation, without prospects. And there are autan information about quantum physics that bases conspiracy theories. Google say, but not learn. A teacher yes. A robot can replace lawyers who help a lawyer, but not the lawyer who pleads. The moocs, poocs and other openclassroom, that’s great. Especially if there is a human in it …
The impact will be felt on the middle classes, through occupations requiring some use of the brain. But not all. Those where it is relevant to centralize a large mass of information to render its parts: lawyers, bailiffs, notaries, teacher, doctors, consultants, journalists. Other businesses affected by the robot will be those that include any repetitive task: the carriers, vendors, dockers, laborers, masons, architects, minor factors. In fact, robots can replace the entire secondary hierarchy of a trade. Much more than cleaning or sports trainers. The middle class will be affected and this is what scares. But it may also be what will allow to enrich the work … From restore meaning.
4 / Technology drift can destroy us …
Let us be clear, I am frankly pro-Robot. I loved my meetings with Nao, Pepper or Tiki. I watch with delight the exploits of Buddy, Kompaï or Parot. I am amazed at the performance Surena 3, Iranian robot, or innovations of the Italian team of the Italian Institute of Technology (IIT — Genoa), on the design of biodegradable robots. And I remain fascinated by the Atlas building (except to pass the broom …). But I see that, so far, if man is capable of imagining, designing, creating, producing, lighting, use, turn off and destroy machine, not the reverse. Humanity went to the moon on its own. Not on a rocket could decide to go alone on Mars …
In 2015, the figures took place, initially against the killer robots, and more broadly on the dangers of AI. Physicist Stephen Hawking believes that once the AI will surpass human intelligence, it could threaten the existence of human civilization. Elon Musk, great AI user for digital money, the private spaceflight and electric cars, has expressed similar concerns. Yet it seems to me that the AI will be what we will do, and exceptional value. Today, the nuclear risk is based primarily on our Korean friend, not an AI error. And despite all our technology, a small group of determined men can reach us in the heart of our capital.
Again, our designs are guided by the presentation made to the AI: the typical futuristic movie pits a small group of individuals, for the control of AI. Or AI fighting against humans for world domination (Terminator, Matrix, I-Robot). And when robots or AI are “nice”, it’s in a cartoon (Wall-E). But that’s not the way the AI is being integrated in the world today; it is not in one or two powerful hands; it is in 1 or 2 billion hands. A child in Africa with smartphone has a more intelligent access to knowledge that the President of the United States 20 years ago. As the AI continues to become smarter, its use will grow. And specifically, the mental capacities of each will be strengthened. Provided knowing where to look and how to use it.
There are strategies to keep AI under control technologies. Biotechnology has twenty years ahead of the AI. In 1975 the potential dangers of the work on DNA were evaluated. The guidelines resulting worked very well and there were no major problems, accidental or intentional, the last 40 years. And now we see major advances in medical treatments. This model can be used for an ethic of AI, which dates back to Asimov’s laws.
Ultimately, the most important approach to reduce potential risks of AI is to work on our human governance and social institutions. We’re already a human machine civilization. The best way to avoid destructive conflict in the future is to continue growing our social ideals, which has already greatly reduced violence. This is the meaning reflections on universal income, the labor code or status of future workers.
The AI now progressing in the diagnosis of diseases, research drugs, the development of clean, renewable energy, education, assistance to disabled people … We can, in the coming years, attending great progress in the fight against the major challenges facing humanity, climate, poverty, malnutrition. And unemployment.
We will probably see the man and the machine alongside one another rather than against: skills focused on innovation, creativity and social skills will be essential to stand out from the crowd of humans and machines and will be required for the maintenance of computers and robots (yes, the machines do not fall ill, but there are bugs or they break). Make no mistake: our current job probably will not exist in 30 years. But is that a bad thing? In our modern societies, have machines that perform tasks without interest and tiring, will allow us to focus on the added value, both in the business, for business development, especially in our own lives, to give meaning to live together.
So I suggest you project yourself to imagine. Imagine not being stuck in traffic jams in the morning. Imagine you adapt your current workload to your lifestyle, rather than the reverse. Working with the world, from any place, at the time that suits you, according to your desires, your needs, your family issues. It would probably much more rewarding that these daily 8/00 in a metro-work-sleep soothing which serves as our living environment for 30 glorious. This is what can bring automation and robotics.
The time when, every day, for 40 years, in the same building, they went up the same stairs to get into the same office, or the next office after promotion, before receiving a nice watch, a beautiful reproduction of an idyllic landscape and a nice speech retracing our wonderful journey that led us young deputy director of the branch office of the authority responsible for internal controls of the audit Office of things checked. Or specializing in debunking to foreman screwing chain worker. Or cashier at WallMart in chief fund to Target. We must adapt to this new, global world and local terribly: innovate to create growth and to create jobs that robots can not do. This relatively easy, provided you adapt to pivot when things do not go our way.
Crafts, repetitive, sometimes quite skilled, will not withstand the robot. Others will develop. Still others create. Some sectors, despite the repetitive aspect, will be spared (security, nuclear), others more or less affected (medicine, journalism, …). Functions requiring discernment, good sense of responsiveness to particular cases, short what can not artificialize, will persist and grow. In the language of the economist Michael Volle, we can estimate that the need for labor will decrease while that in “brain work” will continue.
I know, the unknown is frightening, and it is in my opinion the main reason why we are spreading fear, mostly in the West (eg Africa sees robotics as a way to act in on equal terms with the rest of the world). But we must think differently. Fear does not avoid danger, they say. Understand what is underlying this movement, both positively and negatively. And take control of our lives and our destiny.
Think, innovate, prepare the machines. Use them to improve our daily. It is the purpose of automation. In this context, yes, robots and other AI will be asked to do more and to replace us in part. To compensate us. But confidence, unique ideas, creativity, are areas where the machines will never be able to compete with us. And that will be the foundation of our future activities. It is estimated that as of 2020, the world will produce 220,000 robots ready to replace us. With nearly 3.6 billion people of working age and whose job is likely to be replaced by a robot, it would take about 16,000 years to do it. Not on the species still exists. So what exactly are we afraid?
Task automation, robotics, technological progress, will reduce the work of man. And this will be to produce more or create more, or to do nothing … This is the company to think about it, and to give birth to the debate, rather than trying to ban or scare.
I see no AI today, no robot (except perhaps StarWar’s C3PO) capable of autonomous tasks without relationships between them, such as serving a meal, make sure the protocol and make coffee. And frankly if humanity is unable to resist the machines that can not even make coffee, we fell very low…