- As robots become more capable, they replace humans in the labour force.
- In spite of their limitations, they will take over many skilled jobs in the coming decades.
- Unless they generate more jobs, they will lead to mass unemployment.
- Employment is the basis of our society, so robots may push us to a post-capitalist society.
A modern factory is operated by a man and a dog. The man’s job is to feed the dog. The dog’s job is to keep the man away from the machines.
It’s an oft-repeated joke about the replacement with human labour with robots. The ‘developed’ world has seen many unskilled jobs replaced by computers in the last two decades, with varying degrees of success. Few outside the industry would notice the difference between a car built by humans and by robots, but I’ve never heard of anyone celebrating the replacement of switchboard operators with machines.
The robots are coming
As my previous post on artificial intelligence described, computers are becoming more capable every year, and they are encroaching on what we used to consider the preserve of the human being. In a follow up to the programme that I based the previous post around, BBC Radio 4’s Analysis produced a further podcast about the potential of computers to replace skilled labour. There’s a further, and briefer, discussion of what computers may be able to do in Humans Need Not Apply by CGP Grey:
I’m approaching the subject partly as a person who expects to live in the future being discussed in the programme, but also as a science fiction writer interested in what the world of the future might look like.
The Analysis programme opens with a discussion of where robots may replace people. Intellectual property lawyer Gregor Pryor describes how ‘discovery’, the process of examining an opposing litigant’s documentation for the salient points, can be carried out by a properly programmed computer.The advantage to a law firm is clear. The software carries out the task much faster than the junior lawyer it would otherwise be assigned to, and does not expect a salary, annual leave, sick pay or any of the other costs of employing a human being. The advantage to the legal profession is less clear, as it means less jobs for lawyers.
The algorithmic solution
Andrew McAfee of MIT’s Initiative on the Digital Economy discusses the potential for computers to replace doctors: ‘what a doctor does … is basically an exercise in pattern recognition…Cutting edge software is better, it’s superhuman, at pattern recognition’. In other words, the doctor’s job can be reduced to an algorithm, and computers are very good at applying algorithms.
McAfee’s assertion raises several objections that are not explored in the podcast. The first will be pointed out by any doctor reading this: most of a doctor’s job that can be reduced to an algorithm already is, but doctors still have plenty to do. Doctors often base their diagnoses on tests that they have little understanding of, but they have a clear understanding of their meanings and the actions they indicate. Even if a computer that could physically examine a patient becomes available tomorrow, its impeccable logic would mislead it as soon as it was faced with a mental health problem.
Lady Lovelace’s objection
Algorithms used in discovery of legal documents or medical diagnostics are designed to replicate the decision making process of a human who makes good decisions. As Humans Need Not Apply explains, a robot can learn by observing the results of a process performed correctly and developing a system to reach the same goal. It can only learn if someone has already done it correctly, so human expertise is still required to train the robot.If humans were not involved in the decisions at all, there would be no one to update the algorithms or even to recognise when they need updating. A diagnostic algorithm would be out of date as soon as a new virus appeared on the scene. Similarly, if a litigant knows what algorithm their opponent will use in the discovery of their documents, they may word the parts they don’t want their opponent to notice in a way that will escape the notice of the algorithm.
The limits of algorithms were recognised even as Charles Babbage developed the concept of his ‘analytical engine’. In 1840, he gave a seminar on it at the University of Turin, which inspired a paper by Luigi Menabrea. The paper was translated into English by Babbage’s friend and fellow mathematician, Ada Lovelace. Lovelace was so fascinated by the analytical engine that her notes take up more space than the paper. They famously include a process for calculating Bernoulli numbers, now acknowledged as the first computer program. Unfortunately, Babbage never built the analytical engine so the program was never tested.
Lovelace included the qualification that ‘the Analytical Engine has no pretensions whatever to originate anything. It can do whatever we know how to order it to perform. It can follow analysis; but it has no power of anticipating any analytical relations or truths’. In Alan Turing’s seminal paper postulating what we now call the Turing test, he broadened ‘Lady Lovelace’s objection’ to label the argument that a computer cannot create anything new.
Lady Lovelace’s objection still stands. In the Analysis podcast, it is repeated by policy analyst and robotics entrepreneur Pippa Malmgren:My personal view is that a robot will never be able to engage in creativity. It can just follow an algorithm that a human being has designed and given it. Therefore it can’t make creative leaps. And that’s where real innovation comes from, from the creativity of the human mind and I just don’t think we should underestimate that.
A further objection to replacing human judgement is that working through a process is often necessary to understand it, and the experience gained by that is necessary to refine and design algorithms.
For example, the pocket calculator has replaced the need for most of us to understand basic arithmetic. It saves us a considerable amount of time and makes us less likely to make mistakes. Yet that time and accuracy comes at the cost of an intuitive understanding of arithmetic, which can only be achieved by using it. It’s not something that most of us miss, but can someone ever learn to be a mathematician without that familiarity with numbers? Similarly, could a lawyer like Pryor really understand the nuances of a case if no one on his team has read the relevant documents, preferring to rely on bullet points produced by a computerised algorithm?There is a danger that if professions such as medicine and law become too dependent on algorithms, the processes behind them will come to be seen as ‘black boxes’. Few people will understand how the algorithm produces the results it sets before them, and even fewer will understand why it works in the way it does. There will be an erosion of human expertise, which will stifle innovation because that expertise is required to develop new algorithms, or to recognise that the algorithm needs updating. An algorithm will not adapt to the appearance of new viruses or evolving legal writing styles unless programmed to.
Machines eat jobs
Whatever their limitations of the approach, computers have been replacing jobs for several decades and they are not about to stop. The question is where it leaves the rest of us, limited as we are by our biological minds. If automation gets work done faster and cheaper than a human, an employer will welcome it far more than the employee it makes superfluous. John Steinbeck’s The Grapes of Wrath gave us an unforgettable portrayal of sharecroppers who lost their homes and their work to the new innovation of the tractor.Britain’s largest container port is in the small town of Felixstowe, which replaced the major port cities of Bristol, London and Liverpool soon after it was opened in 1967. One of the main reasons for its near monopoly is that when containers were introduced, allowing one man with a crane to move in an hour as much as a gang of stevedores could move in a week, the dockers’ unions objected to the unavoidable redundancies. Industrial action was circumvented by building a new container port and recruiting new workers who were not members of the unions. Britain cut the price of imports and exports at the cost of widespread unemployment in the old port cities.
The fates of the sharecroppers and the stevedores illustrate a common trend in high income countries over the last hundred years. Unskilled jobs have been either replaced with machines or outsourced to low wage countries. Technology has facilitated outsourcing as well as mechanisation. Satellite communication has cut the cost of international communication so far that call centres can be located on different continents to the customers they are supposed to service.
The economy adaptsThe immediate consequences of job replacement can be deeply unpleasant, as The Grapes of Wrath shows so vividly. However, the longer term consequences have often been more positive. The descendants of the American sharecroppers and British stevedores are far better off than their ancestors. The process that led to their jobs being mechanised also drove the enormous economic improvements of the second half of the 20th century. The increasing wealth circulating in the British and American economies increased the demand on the service sector, opening up a new set of jobs for the unskilled worker. Retail assistants and telephone operators may be insecure and poorly paid, but they are far better off than stevedores or sharecroppers who lived lives of backbreaking labour in exchange for a pittance.
Further, service sector jobs are unlikely to be replaced by machines as people pay for interaction with other people. Gavin Kelly, CEO of the Resolution Foundation think tank that studies living standards, points out that an aging population needs carers and that while automated carers are being developed in Japan, most people would rather be cared for by another human being than a machine. As Kelly puts it, ‘I’ve yet the machine which is going to cut someone’s hair or look after my or your mum’.
The robots have been coming for years
At this point, it’s easy to roll the eyes at the memory of the heady days when workplaces were first being computerised. Think tanks were telling us that the computer was going to be such an effective labour saving device that our biggest problem would be working out what to do with all the leisure time we would have.
Needless to say, it didn’t work that way. Computerisation did increase productivity but in doing so, it devalued human labour. In a report titled Technology at Work, Carl Frey and Michael Osborne of the Oxford Martin School show while we’ve been producing more units of GDP per hour we work every year since 1970, the pay for each hour worked has not been keeping pace since 1995 (Fig. 4, p.10). When computers helped us work faster, we simply found ourselves with more work. Meanwhile, a host of new job descriptions appeared for the people who develop software and maintain the computers.
Will the robots take our jobs or create new ones?
Malgren’s view is that far from making humans redundant, more capable computers will create more jobs:
All I know is from my work in the field, what I see, is more and more people being hired into companies that are engaged in the creation of robotics. There’s just no way to prove it yet but I think we’re going to find that like all technological innovation, we’ll end up creating more opportunities ultimately than destroying them.
The capitalist computer
But what if CPG Grey is right when he says ‘this time is different’? What if Malgren’s belief that technology will create more jobs than it destroys is as misplaced as Grey’s complacent horse contemplating the arrival of the automobile? As someone who would have to live in the resulting economy, I find that frightening. As a science fiction writer, I find it exciting.
Kelly’s view is that there will always be jobs that require the human touch, but those jobs are typically among the poorer paid jobs in the economy. It’s the better paid, more skilled jobs that are at risk. If you’re reading this and
thinking you’re not a doctor or a lawyer so it’s not your concern, think again. The service sector is only viable if there are enough people with money to spend on it. No one will be employed to care for an elderly person whose younger relatives are unemployed, as they will have time on their hands but no money. No one will be employed to make lattes or even flip burgers if there aren’t a reasonable number of people with money to spend on them.
In fact, the trend has already begun. Frey and Osborne’s report shows that between 1993 and 2006, Europe and the USA saw jobs move away from middle-wage jobs toward higher and lower ends of the wage spectrum (Fig. 9, p.20). We’ve heard a lot of talk about the ‘squeezed middle’ in the media, and Frey and Osborne show us what that means. A large gap is appearing in the middle of the salary spectrum, right where the computers postulated by the Analysis podcast are likely to find their niche.
Will the robots impose a post-capitalist dystopia?In a capitalist economy that has no need for workers, there will be no need for the few people who own the capital to share it with the majority who don’t. David Graeber, professor of anthropology at the London School of Economics, is as worried as we might expect from the man credited with coining the slogan, ‘we are the 99%’:
Will capitalism survive? I suspect not. So it strikes me that we are going to move to something else. I’m a long term thinker and one of my greatest worries that that whatever the next thing is, it’s going to be even worse. It could be some new form of feudalism or fascism or technofascism of some kind. That could easily happen.
When one of the staunchest critics of the modern capitalist system is worried about what might replace it, it’s time to be very worried indeed.
Will the robots give us a post-capitalist utopia?
If computers threaten the capitalist system, we have two options. One is to form a neo-Luddite movement to smash the machines that threaten our jobs. Another is to build a system that can accommodate the machines without disenfranchising humans. Given that the original Luddites were as unsuccessful as every other anti-technology movement in history, the latter is the more realistic option. Federico Pistono, author of Robots Will Steal Your Job, But That’s OK: How to Survive the Economic Collapse and Be Happy, suggests that ‘outsourcing intellectual labour’ may be an opportunity to re-evaluate the economy we live in and our place in it:
We haven’t asked ourselves what is the purpose of, for example, work. It should be to increase your standard of living so you can do the things you like.In our current system, it sounds like pie in the sky. If we don’t work, we don’t get paid and if we don’t have money, we can’t function in a capitalist economy. Yet as Graeber pointed out, if machines replace humans as the labour that drives the economy, the capitalist system will have already broken down.
Pistono suggests that rather than an economy in which we exchange time and labour for a little piece of the economy in the form of currency, an automated economy would generate enough wealth to allow us all to live in comfort. It’s a utopian vision that will require some significant problems to be addressed.
How to prevent the owners of capital from subverting it to their own means? How to reward the people it will need to manage, defend and police the system without giving them draconian powers or the opportunity for corruption? How to maintain a healthy economy in a world where national economies will still be competing with one another?
From the welfare state to the welfare society
Many countries have already made the first step, acknowledging that the economy should provide a social safety net. I’m fortunate enough to live in the UK where the state provides housing and benefits for the unemployed, healthcare for the ill and a pension for the elderly. It’s not as generous as many other European countries, and it is in danger from a neoliberal ideology that exalts the free market.Pistono’s vision suggests that rather than lowering the social safety net, the automated economy will raise it so high that it will be the main source of income for many people.
Whether we’re headed for Graeber’s post-capitalist dystopia, Pistono’s leisurely utopia, or Malgren’s status quo with more robots, we’ve already started the journey. Frey and Osborne predict that in ‘developed’ countries, 30-60% of jobs will be replaced by computers in the next 20 years (Fig. 49, p. 62). They probably won’t stop there.
It’s no longer true to say the robots are coming. The robots are already here. It’s up to us to decide what we’re going to do about it.