A strange vision is gaining traction in the developed world: a vision of a future economy so dramatically changed by technology that it is beyond recognition. This is not the apocalyptic prophecy of a real-world Skynet, turning computerised gadgets and artificial intelligence systems against the human population, but a growing fear of automation replacing people in an increasingly robotised world of work.
While it is plain to see how machines have improved humanity’s lot in the past despite putting their overlords out of work, some argue that this time, it’s different. No doubt, sophisticated IT and AI are likely to consign more jobs to the bin of history than ever before. But some elements of this new techno-pessimism are worryingly short-sighted, and they must be weeded out before they gain more popularity in business and political circles.
The Invisible (Steel) Hand
The threat that automation poses to today’s jobs is very real. What used to be seen as a somewhat unquantifiable aspect of the global economic future is quickly becoming the subject of rigorous study. Over the last year, for instance, big organisations like McKinsey, Deloitte and Merrill Lynch have poured resources into analysing the likely effects of automation on the workplace.
McKinsey analysed 800 different occupations in the US, assessing the feasibility of automating the various activities involved in each profession and across sectors. The results are striking: they estimate that 27% of activities in the education sector have the potential to be automated with already existing technologies, whilst in jobs involving physical activity and machine operation in a “predictable environment” the figure rises to 78%.
The McKinsey report argues that analysing work activities foremost, rather than occupations as a whole, gives the most accurate picture of the technical feasibility of automation.
Looked at from this perspective, it is difficult to see why entire occupations should be wiped out in droves by automation. Many jobs today involve a number of different types of activities. In those, machines taking over the more mundane tasks means that humans are freed up to work on the more complex tasks for which they are still required, and thus become more productive – or to spend less time working and still produce the same output. Robots have never made their meat-bag operators redundant in these kinds of occupations, and there seems little reason to think that they ever will.
A New Techno-Pessimism?
Policymakers, of course, are concerned about those working jobs which do stand to become entirely automated. These jobs are those which involve more or less a single automatable task, or those in which all tasks have the potential to become automated in a very short space of time.
Though they would be quick to distance themselves from the Luddites who revolted against the rapid changes of the Industrial Revolution, the same principles are involved. After all, the redundancy of human labour is nothing new. The replacement of humans in manufacturing continues at a rocketing pace: the supply of robots to industry alone rose from 110,000 in 2005 to 230,000 in 2014.
There are many sides to this new techno-pessimism. One version more obviously involves a resurgence of the fears which drove 19th Century textile workers to wreck the newfangled contraptions which were displacing them en masse. The difference today is that automation is set to threaten more well-paid, skilled jobs. Now, jobs occupied by the middle classes are under threat and those from that category find themselves much less able to face so-called “technological unemployment” with the aloof perspective that economic history takes on the Luddites. The sense of panic, it seems, spreads more effectively from the middle classes into political and academic spheres.
Some are worried that the result will be a bifurcation in the job market: that the share of jobs will fork off between a smaller and smaller number of high-paid executives, managers, and software programmers, and the rest of low-paid, low-status jobs.
This is based on the expectation that sophisticated software will take over mid-level jobs in the service sector, wiping out the middle classes in the process. Indeed, automation is already emerging in fields like law, accounting and finance.
Those who insist that this bifurcation is reality often refer to the recent increase in lower-wage jobs. Since the 2008 recession, for example, the greatest job growth in the economy has come in the low-paying service sector.
Leaving aside the myriad of other things that can explain the shift, is this, in fact, good evidence that automation will take it further? If low-paid service jobs are able to boom at a time when new technology has a quickly expanding role in them – think of software’s replacement of human roles in restaurants, banks, taxis and so on – then this trend says little about what will happen to middle and high-income jobs become threatened by more sophisticated technology.
The techno-pessimists’ predictions make assumptions that do not hold up in the real world. They presume that, by presenting a cheaper alternative to their human counterparts, automated systems and machines will mean fewer jobs in a given occupation, and will mean more money only for the businesses that own them.
The trouble is, this doesn’t happen in reality. Take, for example, the use of sewing machines in places like Bangladesh. It makes it cheaper to produce clothes – it does not mean that fewer people are needed to satisfy today’s demand for clothes. Each item requires far fewer human hours than in the days of hand-sewn garments, and labour is paid for not per person but by the amount of time spent working.
The techno- pessimist’s attitude makes the error of focusing purely on the value of the seamstress’ work before the machine’s introduction compared to afterwards. It ignores the huge number of people who were able to find work due to the fewer technical skills required to produce garments and the immensely greater number of garments needing to be sewn in light of their lower cost of production. And it ignores that this work is more gainful than what they had before (or did not have, in the case of the unemployed) since they chose to join the profession. Many more workers in Bangladesh are now employed in the textile industry as a result of the sewing machine, and for more than they earned before.
Reasons To Be Cheerful
The same thing goes in this new era of automation. Much has been made, for instance, about how computerised systems are replacing the human role in legal work. As software gets better and better, the lawyers of today will likely find their pay dropping (something which the general public might find easier to swallow than the deflated wages of subcontinental seamstresses). At the same time, though, it will take much less to become a fully-trained lawyer as computers take care of more and more of what today’s lawyers have to do, and legal work will become far cheaper by the unit.
The techno-pessimist fears hordes of barristers on benefits; the reality will be greater access to legal work for poorer individuals and smaller businesses, and a more productive and numerous workforce in the legal profession. Indeed, a 2015 study by economists at Deloitte found that knowledge-intensive sectors such as accountancy have already seen simultaneous rises in productivity and employment due to technology.
Now, it may still be the case that advances in automation will reward those at the top more than it will benefit those on the bottom and middle rungs of the job ladder. But there is nothing to suggest that those lower down will not be better off in absolute terms. And even a relative advantage to those at the top is incredibly difficult to predict given the sheer number of variables, both in the wider economy and the technologies themselves, that determine whether the machines’ owners will gain more than the rest of society.
The more rational side of the techno-pessimist trend focuses on the rate at which technology is growing. It seems that the rise of AI systems and the like mean that activities previously requiring human input will enter the machines’ dominion at such speed that they will make humans redundant faster than they can cope with.
Although current innovations typically take one task out of employees’ hands at a time, allowing people to be used more productively elsewhere, what about when this happens so quickly that they lose their jobs altogether?
Historically people’s skill-sets have changed dramatically, according to the job market’s evolution in line with technology. But, just as it was short-sighted to expect displaced 19th-century workers to easily find work elsewhere, it would be obtusely Victorian to demand that people in the service sector pick up managerial and computing skills before they feel the pain of unemployment.
It takes time for the opportunities that technology offers to filter through to those who have been displaced by it. The Luddites would have had to suffer lower living standards until their training or wage inflation brought them up to the level that they were before the machines arrived.
As a result, attention has been heaped on the crucial role of education. The worry is that those set to be replaced by automated technologies might not have the time or the capital to retrain or re-educate themselves in time to meet the demands of a rapidly evolving labour market.
Automation and The Skills Dilemma
The biggest unknown, though, is what information to base this on. With automation advancing in ways that the average person, and indeed most policy makers, are increasingly incapable of keeping up with, who is to know where best to pour educational resources into?
Governments have not been particularly good at this, with many Western school systems arriving rather late to the game when it comes to teaching computing and programming. There are plenty of examples of businesses being caught out too: this year Google announced plans to sell Boston Dynamics – famous for developing impressively life-like robotic systems like Big Dog – since it has not performed well enough financially. Meanwhile, its driverless car project, the darling of the tech press, has not made anywhere near the kind of progress expected for actual implementation.
It seems that people replaced by machines will be forced to accept periods of redundancy and lower wages until they manage to adapt. The worst case scenario is that a generational gap emerges whereby individuals never quite manage to recoup the lost value of their expertise, and only their children end up with skills suited to the automated world.
DIY In Literacy
One should not assume, though, that adapting through education and retraining will be a winner-takes-all game in which a lucky few catch onto the latest trend in programming or robotics while the rest are condemned to a lifetime of unemployment. There is simply no telling what new professions will be created by technology itself, and which existing ones will find their ranks swelling as technology continues to make necessities cheaper and people find other things to spend their income on. For instance, the two fastest-growing occupations of the last five years in the UK are Zumba instructors and personal trainers: two services that most people didn’t remotely want to spend their money on a mere few years ago.
And importantly, today’s children are growing up with far more transferable skills like reading and writing than their Luddite ancestors who had almost nothing to offer but their weaving, and so are better prepared to adapt. What’s more, the nature of education itself could well change as a result of automation. The Global Learning X Prize is offering $15m to whoever develops the best software enabling children to teach themselves literacy and arithmetic. Big things are on the horizon. Perhaps fittingly, it may be that what prepares the world best for the advent of technology will be technology itself.
History As Usual
Few seem to be challenging the huge gains in living standards that humanity owes to automation so far. The Deloitte economists’ study analysed 140 years of data on census results in Britain and concluded that machines have created far more jobs than they have destroyed. Ultimately, there is nothing to suggest that the same process of creative destruction will continue for the next hundred years, but there is nothing to suggest that it won’t stop accelerating, either. The question is how to keep up.