Did you know that it’s more difficult for a robot to walk by chewing gum than playing chess? #sortlechampagne That’s almost good news, is not it? Because the almost daily display of the power and precision of artificial intelligence algorithms often has the effect of a cold shower on the idea that we have of our own intelligence.
Just think of AlphaGo. This gaming program, driven from 30 million games played by humans, managed to beat the world champion 4 games against 1 … before being vitrified by another program, AlphaGo Zero. Powerful machines.
And if these machines came to pass us? If they stole our jobs?
A report published by the Organization for Economic Co-operation and Development (OECD) in early April somewhat alleviates this fear. While researchers at Oxford University claimed in 2013 that 47% of jobs were at risk of being automated, the OECD is talking about an average of 14% of jobs. That’s still 66 million people at risk of losing their livelihood.
Truck drivers are often cited as examples of risky jobs. Except that, for the moment, autonomous trucks have much less difficulty driving on empty and straight motorways in Arizona than in a busy downtown Quebec city at the end of the afternoon during a snowstorm.
It is still too early to bully. For the moment, the more repetitive the task definition is or the processing of large sets of structured data, the more it is at risk.
Paradoxically, many experts say that low-paying jobs that require human qualities (such as empathy) could be spared. In summary, a legal assistant is at greater risk than a beneficiary attendant. The explanation lies perhaps in this news, released last week, where we see robots succeed in painstakingly assembling an IKEA chair .
The white man and the paradox of Moravec
Technology is never neutral. Never. It’s as solid as Kanye West’s ego problem.
“In 3 to 8 years, we will have a machine with the general intelligence of a human being,” said Marvin Minsky, one of the founders of the science of artificial intelligence, in Life magazine in 1970. Almost 50 years later, several experts claim, on the contrary, that it may even be a technologically impossible thing.
But why did Minsky, that intelligent man, so badly deceive himself? Notwithstanding his genius, he belongs to one of the most favored socio-demographic niches in the history of humanity: the white American man. At the founding conference of the field of artificial intelligence, at Dartmouth College, New Hampshire in 1956, there were only white men. The only woman present was Marvin Minsky’s wife, Gloria.
And what was the value of educated white men of that time? Equations, calculation and … failures. The ultimate stage of artificial intelligence was to beat a human in this game. This will be successful in 1997.
The biologist Henri Laborit has often compared the phenomena of our consciousness, what we call reason, and all that flows from it (language, self-representation, calculation, inner dialogue, etc.) to a sort of an elusive foam that surfaces on the surface of the ocean of our unconscious.
From an evolutionary point of view, he is right. All of this has happened very recently in our evolution, a few hundred thousand years at most. And that’s where computers often plant us; when it comes to our most recent learning.
The thesis of the roboticist Hans Moravec, called the paradox of Moravec, is based precisely on this point: screwing eight pieces of wood together is more difficult for a robot than managing 80 billion trillions of possibilities in the game of go.
I’m walking so I’m
The human being easily achieves apparently simple operations, but which are actually of bewildering complexity, such as walking and chewing gum at the same time. And during this time, we have difficulty playing chess. All this is highly logical, as Spock would say.
Chewing, swallowing without choking, maintaining one’s balance, advancing in urban environments filled with unexpected and threatening matters for us to walk health. We even do it to rest. Why? Because behind these behaviors, there are billions of years of evolution. Why are our feet so skillful? Why can we spot obstacles (holes, animal dung, puddles of coffee) and threats without even realizing it? Developments.
Do not forget that the pair of eyes with which you read this text (and all the neural apparatus that comes with it) has been perfecting since … 600 million years ago! The first traces of writing and calculations appeared 5000 years ago! For Hans Moravec, the ease with which we learn, for example, to ride a bike, and the difficulty we have to concentrate on reading texts is a matter of time.
Nobody is a 100 watts
One of the most admirable qualities of the human brain, in addition to its ability to create memes from Yves Corbeil’s photos, is its energy efficiency. It does not always seem, but we are energy efficient people.
In a paper published in 2012 in Scientific American magazine , journalist Ferris Jabr estimated the average consumption of a human brain. Imagine 1400 cubic centimeters of fat matter made up of about 86 billion neurons connected by something like 1000 billion connections. Total amount of electricity consumed: 12.6 watts. Have you ever tried reading with a 15 watt lamp?
The team behind AlphaGo had a hundred scientists. AlphaGo’s “engine” had a whopping 1,200 processors and 176 graphics processors. Power consumption: 1 megawatt, the equivalent of the consumption, per hour, of several hundred houses. And do not try to ask AlphaGo to tell you the weather.
The next time a computer will beat you in chess, comfort yourself by telling yourself that walking and an excellent balloon gum are, for several more years, specificities that can not be more humane.
In closing, quickly, here are the secrets of a happy couple: listening, empathizing and never trying to assemble an IKEA furniture together. Evolution has not yet prepared us for that.
John Gilmore was a reporter for Techno Secrets, before becoming the lead editor. He has over fifty bylines and has reported on countless incidents around Anchorage. John studies chemistry and history at the University of Alaska Fairbanks where he currently is in his senior year.