When in 1940 Isaac Asimov wrote the first in a collection of
stories which was later published as “I, Robot” he could barely have dreamed
how far we would advance in the subsequent decades. One of the biggest economic
and social challenges of the coming years will be how to deal with the rapid
surge in automation which threatens to destroy jobs on an unprecedented scale.
Will we find alternative means of employment? Or will we be cast aside as the
second machine age (as Brynjolfsson and McCaffee so memorably called it) gains
pace?
I have long been fascinated by the application of computers to automate routine tasks and recall a lecture I attended more than 30 years ago when I was introduced to the concept of artificial intelligence. AI had first surfaced in the 1950s when computer applications were developed which could perform fairly routine tasks very well – and indeed, in some cases better than humans. But the difficulties in getting computers to think and act like humans had been underestimated and by the mid-1970s interest started to languish. By the time I became acquainted with the subject, interest had been rekindled by so-called expert systems which utilised programs based on ‘if-then’ rules rather than highly structured procedural programming. A new wave of AI was kicked off in the 1990s by improvements in computing power and advances in deep learning technology, which is a statistical technique designed to allow systems to learn from observational data. By 1997, the world chess champion, Garry Kasparov, was being beaten by a machine and by 2011 IBM’s Watson computer was able to beat human contestants in the TV general knowledge quiz Jeopardy!
There are a number of different strands to the technological revolution incorporating software and hardware innovations which have fuelled concerns that many of us will be replaced by a machine in the not-too-distant future. Indeed, much of the talk today is of how far technology will advance in the coming years with apparently simple skills such as driving now able to be replicated with a high degree of accuracy by machines, That said, a reasonable degree of accuracy is not yet good enough – machines will have to do as well as, or better than, humans in order to displace them. But with 3.5 million people in the US alone employed driving trucks, that is a lot of workers who could potentially be displaced. It is against this backdrop that Bill Gates recently suggested that a tax should be levied on robots in order to safeguard the jobs of humans.
One of the motivating notions behind the idea is that those with access to capital, who employ robots in the first place, would be enriched at the expense of those whose jobs are displaced. The idea is thus to levy a tax in order to reduce inequality. Another motivation is that since it takes decades for the full impact of a major technological revolution to work through, the imposition of a tax will slow the widespread adoption of automation rather than stop it in its tracks, thus giving the workforce time to adjust.
Whilst these are laudable aims, there are serious difficulties involved in imposing a robot tax. The most obvious problem is that it impedes innovation. Imagine where we would be today if the automobile had been taxed to the hilt in order to preserve the jobs of horse carriage drivers. But the sheer range of jobs which can be automated suggests that new technology threatens to displace workers not seen on a scale since the start of the industrial revolution – and even then, displaced agriculture workers were able to find jobs in the rapidly growing urban areas. A bigger problem is that it is difficult to differentiate between technologies that substitute for labour and those which complement it. The introduction of the personal computer in the 1980s had a dramatic impact on the world of work but this has not prevented the numbers of people in employment in the US and UK reaching record highs. A combination of human effort and new technology has facilitated a whole new range of jobs that were previously unimaginable.
So if taxing robots is not necessarily practical, is there anything else we could do? The blogger and economist Noah Smith suggests introducing a wage subsidy for low-paid workers – perhaps by cutting payroll taxes – which will raise the cost advantages of human labour versus capital. Another suggestion might be to distribute capital income more widely by using tax revenue to buy assets (real and/or financial) and distribute the profits to the wider population. The idea is that this gives citizens a stake in society and allows them to share in the profits generated by the new robot labour force.
Although this idea may be anathema to those who believe in ever smaller government, and that the displaced will have to find new ways to compete in a new technological world, the reality is that the pace of change may be too rapid for many. Indeed, unless we start to think about ways to cope with the problem today, modern industrial societies could be overtaken by events. And if we think people are dissatisfied today, wait until they are replaced by machines, with no hope of finding a new job. It could make the Trump/Brexit revolution seem like a picnic.
I have long been fascinated by the application of computers to automate routine tasks and recall a lecture I attended more than 30 years ago when I was introduced to the concept of artificial intelligence. AI had first surfaced in the 1950s when computer applications were developed which could perform fairly routine tasks very well – and indeed, in some cases better than humans. But the difficulties in getting computers to think and act like humans had been underestimated and by the mid-1970s interest started to languish. By the time I became acquainted with the subject, interest had been rekindled by so-called expert systems which utilised programs based on ‘if-then’ rules rather than highly structured procedural programming. A new wave of AI was kicked off in the 1990s by improvements in computing power and advances in deep learning technology, which is a statistical technique designed to allow systems to learn from observational data. By 1997, the world chess champion, Garry Kasparov, was being beaten by a machine and by 2011 IBM’s Watson computer was able to beat human contestants in the TV general knowledge quiz Jeopardy!
There are a number of different strands to the technological revolution incorporating software and hardware innovations which have fuelled concerns that many of us will be replaced by a machine in the not-too-distant future. Indeed, much of the talk today is of how far technology will advance in the coming years with apparently simple skills such as driving now able to be replicated with a high degree of accuracy by machines, That said, a reasonable degree of accuracy is not yet good enough – machines will have to do as well as, or better than, humans in order to displace them. But with 3.5 million people in the US alone employed driving trucks, that is a lot of workers who could potentially be displaced. It is against this backdrop that Bill Gates recently suggested that a tax should be levied on robots in order to safeguard the jobs of humans.
One of the motivating notions behind the idea is that those with access to capital, who employ robots in the first place, would be enriched at the expense of those whose jobs are displaced. The idea is thus to levy a tax in order to reduce inequality. Another motivation is that since it takes decades for the full impact of a major technological revolution to work through, the imposition of a tax will slow the widespread adoption of automation rather than stop it in its tracks, thus giving the workforce time to adjust.
Whilst these are laudable aims, there are serious difficulties involved in imposing a robot tax. The most obvious problem is that it impedes innovation. Imagine where we would be today if the automobile had been taxed to the hilt in order to preserve the jobs of horse carriage drivers. But the sheer range of jobs which can be automated suggests that new technology threatens to displace workers not seen on a scale since the start of the industrial revolution – and even then, displaced agriculture workers were able to find jobs in the rapidly growing urban areas. A bigger problem is that it is difficult to differentiate between technologies that substitute for labour and those which complement it. The introduction of the personal computer in the 1980s had a dramatic impact on the world of work but this has not prevented the numbers of people in employment in the US and UK reaching record highs. A combination of human effort and new technology has facilitated a whole new range of jobs that were previously unimaginable.
So if taxing robots is not necessarily practical, is there anything else we could do? The blogger and economist Noah Smith suggests introducing a wage subsidy for low-paid workers – perhaps by cutting payroll taxes – which will raise the cost advantages of human labour versus capital. Another suggestion might be to distribute capital income more widely by using tax revenue to buy assets (real and/or financial) and distribute the profits to the wider population. The idea is that this gives citizens a stake in society and allows them to share in the profits generated by the new robot labour force.
Although this idea may be anathema to those who believe in ever smaller government, and that the displaced will have to find new ways to compete in a new technological world, the reality is that the pace of change may be too rapid for many. Indeed, unless we start to think about ways to cope with the problem today, modern industrial societies could be overtaken by events. And if we think people are dissatisfied today, wait until they are replaced by machines, with no hope of finding a new job. It could make the Trump/Brexit revolution seem like a picnic.