Oncologists, psychologists, bus drivers, financial analysts, personal assistants and sales assistants: machines can do all their jobs. How, in the space of twenty years, have machines succeeded in performing so impressively, which jobs are most under threat and are there still fields in which human beings do better?
In our collective unconscious, a robot is either an omniscient computer, which ends up getting the better of humans like HAL 9000 in the film 2001 Space Odyssey, or is an android capable of feeling emotion and resembling us so closely that it becomes impossible to unmask it, as in another cult film, Blade Runner. We may naively believe that it is not tomorrow that we will come face-to-face with this type of robot, but we would be wrong. Unnoticed by us, they are already everywhere: clearly they do not resemble how they are portrayed in films and books.
Artificial intelligence is omnipresent: a confirmation e-mail received after making an online purchase is obviously not written by a human being but a machine. Heading somewhere by car, it is again a machine which calculates the route depending on traffic jams and road works, and then guides you. When you buy a book via a website, it is artificial intelligence which suggests a list of other books that may interest you.
Jobs yes, but for machines
Many universities and think tanks are studying the consequences of this evolution, particularly in the jobs market. The consultancy, Roland Berger Strategy Consultants, has carried out such research in France and has revealed that 3 million jobs could be entrusted to machines over the next ten years. According to their calculations, 42% of jobs are at risk in this way. Two researchers from Oxford have carried out a similar analysis of employment in the USA and arrived at a relatively similar figure: 47% of jobs could be automated within ten to twenty years.
For a fistful of dollars
How can we explain this meteoric rise of robots in the workplace? The answer is the various technological advances which are behind it. The first of these is computing power. In an article in the November 2014 issue of the magazine Science et Vie, the following interesting comparison was given as an illustration: in 1997, the supercomputer ASCI Red, designed to simulate nuclear tests, occupied the space of a tennis court, performed 1800 billion operations per second and cost 55 million dollars. Ten years later, PS3 achieved the same computing power for a fistful of dollars and took up just a few cubic centimetres.
Machines adapt to the unknown
The second determining factor is big data, or databases which contain masses of information, since everything or virtually everything is now digitised and stored in the Cloud; from there, a connected robot armed with high-performance algorithms can perform miracles when it comes to finding, sorting, analysing and even interpreting information and taking decisions. Recent years have seen significant improvements in robots’ capacity to adapt to unknown environments. “This is a crucial point,” believes Micha Hersch, author of a thesis on cognitive robotics and researcher with UNIL’s Computational Biology unit. “This enables the robot to move on from the assembly line and to react to events in its surroundings, even going as far as to interact with humans or other machines in an open space. As a result of this latest competence, robots can be seen deployed in war zones or in the Google car which drives itself.”
What is worrying is that between 42% and 47% of jobs will be affected in the next ten to twenty years. Writing in Le Temps on 12 December 2014, Olivier Feller, member of the Swiss National Council for Vaud, is concerned about this issue, revealing that, while imminent robotisation is going to create technology sector jobs, “they will far from compensate, either in terms of numbers or skills required, for the jobs which disappear.”
Jobs the robots will steal from us
Are all employees equally under threat? No. Like the manual workers, who have already paid a heavy price for industrialisation followed by automation, it is the tertiary sector which will now be besieged by machines. Carl Frey and Michael Osborne, the two Oxford researchers who analysed the American jobs market, included in their study a list of the fields most at risk. In descending order they cite jobs in transport and logistics via interconnected databases, followed by office and administrative services, goods manufacture – which is set to become even more automated than at present – caring professions and retail checkouts.
As seen above, technological developments are the root cause. However, as Micha Hersch points out, “it is also because the way in which these jobs are carried out has changed enormously. By establishing protocols and procedures for all decision-making, we are getting closer to how a computer functions. In the past, a banker would approve a loan for a house on the basis of what he knew about his customer from a banking relationship spanning years – the figures and salary levels were there but so was intuition. Nowadays, whether it is for loans, portfolio management or another service, a banker often applies protocols decided on by the bank’s hierarchy. With such criteria in place, it is easy to create software which responds with yes or no to questions posed online.”
We are not equal to machines
In the ‘fight’ between humans and machines for a job, not everyone is equal. There are, of course, certain jobs which it is difficult to delegate to machines, such as those involving a high degree of creativity or based on social and human bonds – teachers, social workers, educators, etc. But these are not the sole criteria; the social status of those whose jobs are threatened is an influential variable. It is simple for a computer to take the place of a doctor: it records a patient’s symptoms and tests; the software compares these with its database, establishes a diagnosis and recommends treatment. “Aside from patients most probably demanding to be treated by a human being, it is also highly likely that doctors will be successful in defending their role and will guarantee the long-term future of their jobs. Underground train drivers less so,” believes Micha Hersch.
Are machines a threat to humans?
While researchers and politicians concern themselves with the impact of the robotisation of jobs on the employment market, others are afraid of the impact on humanity full stop. This is the case with the celebrated physicist Stephen Hawking who, during an interview with the BBC, shared his fears about artificial intelligence which could “take off on its own and re-design itself at an ever-increasing rate. Humans, who are limited by slow biological evolution, couldn’t compete.”
Such evolution lies in the realms of fantasy. Laurent Keller, professor in the Department of Ecology and Evolution in the Faculty of Biology and Medicine, has used robots for research into his specialism. “The idea was to improve understanding of collaboration and altruism in ants by studying the importance of family ties in particular and by retracing whether these ties were established from the start. Now that all ants are altruistic, it is impossible to know how it all began.” The robots used for this research have a network of neurons which are subject to and can evolve over the course of generations of selection. In particular, therefore, they have enabled Laurent Keller to go back in time, playing to some degree the role of original ants with and without family ties. But it has also been possible to accelerate the process using computer selection of robots over hundreds of generations. Professor Keller has thus discovered that the “robot ants learn to lie, so as not to have to share food when selected over generations in colonies containing unrelated individuals. In contrast, cooperative and altruistic behaviours rapidly emerge over generations of experimental evolution when the artificial ants are in societies made up of related individuals,” he explains. Professor Keller for one is clearly not ready to be replaced by a robot.
He does, however, continue to make use of machines in his research, in this case a computer, to analyse the movements of hundreds of ants and to distinguish patterns that human intelligence would take years to see.