Robots and capitalism
Can new technology and artificial intelligence open up a new period of expansion for capital? Michael Roberts examines the possibilities and problems
In the last few years, Robert J Gordon, a professor of economics at Northwestern University, has persistently argued against the trendy view of the moment that robots, artificial intelligence (AI) and other ‘disruptive technologies’ are about to launch the global economy into a productivity boost never seen before.
Gordon argues that the arrival of the internet and mobile phones has failed to generate a sustained upturn in the growth of productivity. Output per hour worked in the US grew at a rate of 3% a year in the 10 years up to 1966 - after which the growth rate declined, falling to just 1.2% in the 10 years to the early 1980s. After the launch of the worldwide web, the moving average rose to 2.5% in the 10 years to 2005. But it then fell to just 1% in the decade to 2015.
A decomposition of the sources of growth in productive capacity underlines the point. Over the 10 years up to and including 2015, the average growth of ‘total factor productivity’ in the US - a measure of innovation - was only 0.3% a year. Productivity recently has been growing well below the 2.1% average gains seen over the past 67 years.
So, according to Gordon, the great new innovatory, productivity-enhancing paradigm that is supposedly coming from the digital revolution is actually over already and the future robot/AI explosion will not change that. On the contrary, far from faster economic growth and productivity, the world capitalist economy is slowing down as a product of slower population growth and productivity.
Now Gordon has compiled all his ideas and retorts to those who have disagreed into a new book, The rise and fall of American growth.1 “This book,” Gordon writes in the introduction, “ends by doubting that the standard of living of today’s youths will double that of their parents, unlike the standard of living of each previous generation of Americans back to the late 19th century.”Gordon predicts that innovation will trundle along at the same pace as the last 40 years. Despite the burst of progress of the internet era from the 1990s, total factor productivity - which captures innovation’s contribution to growth - rose over that period at about one-third the pace of the previous five decades.
Gordon reckons that the American workforce will continue to decline, as ageing baby-boomers leave the workforce and women’s labour supply plateaus. And gains in education - an important driver of productivity that expanded sharply in the 20th century - will contribute little. Moreover, the growing concentration of income means that, whatever the growth rate, most of the population will barely share in its fruits. Altogether, Gordon argues, the disposable income of the bottom 99% of the US population, which has expanded about 2% per year since the late 19th century, will increase over the next few decades at a rate little above zero.
It is a grim picture that Gordon paints for the future of US capitalism and thus the world. But is it right? The argument against it is that people are adopting new technologies, including tablets and smartphones, at the swiftest pace seen since the advent of the television. While television arguably detracted from US productivity, today’s advances in technology are generally geared toward greater efficiency at lower costs and so will boost the productivity growth of labour.
Some argue, against Gordon, that statisticians are failing to measure output correctly, partly by failing to capture free services, such as search, which generate vast unmeasured surplus value. But, as Martin Wolf of the FT pointed out recently,
it is not at all clear why statisticians should have suddenly lost their ability to measure the impact of new technologies in the early 2000s. Again, most (past) new technologies have also generated vast unmeasured surplus value. Think of the impact of electric light on the ability to study.2
Nevertheless, balanced against Gordon are a myriad of techno-optimists and economists who reckon that the world is on the brink of a productivity explosion driven by robots, artificial intelligence, genetics and a range of new ‘disruptive technologies’ - disruptive in the sense that traditional jobs and functions are going to disappear, to be replaced by robots and algorithms. The optimists argue that, since the time of Thomas Malthus, eras of depressed expectations like our own have inspired predictions of doom and gloom that were proved wrong when economies turned up a few years down the road.
Kenneth Rogoff of Harvard University pitched in on Gordon’s predictions in a recent article.3 He agreed that there were obstacles to continuing the ‘previous success’ of capitalism. There was environmental degradation; growing inequality within countries; ageing populations that do not work; and the risk of financial crashes. Yet he remained optimistic that capitalism can overcome these challenges. After all,
so far, every prediction in the modern era that mankind’s lot will worsen, from Thomas Malthus to Karl Marx, has turned out to be spectacularly wrong … despite a disconcerting fall in labour’s share of income in recent decades, the long-run picture still defies Marx’s prediction that capitalism would prove immiserating for workers. Living standards around the world continue to rise.
Rogoff continues that technological progress has trumped obstacles to economic growth in the past:
Will each future generation continue to enjoy a better quality of life than its immediate predecessor? In developing countries that have not yet reached the technological frontier, the answer is almost certainly yes. In advanced economies, though the answer should still be yes, the challenges are becoming formidable.
So mainstream economists remain broadly optimistic about the future of capitalism, despite Gordon’s prognostications - not surprisingly.
Gordon’s main opposition comes from professor Joel Mokyr who works with Gordon at Northwestern University. What Gordon fails to account for, professor Mokyr argues,4 is that the information technology revolution and other recent developments have produced mind-blowing tools and techniques, from gene-sequencing machines to computers that analyse mountains of data at blistering speed. This is creating vast new opportunities for innovation, from healthcare to materials technology and beyond: “The tools available to science have been improving at a dazzling rate … I’m not sure how, but the world of technology in 30 to 40 years time will be vastly different than it is today.”
Every day there is a new story in the media about how people and their skills are being, or will soon be replaced by machines and computer software that learns for itself. At this year’s World Economic Forum - the annual meeting of the global elite of bankers, politicians, corporate chiefs and military - in Davos, Switzerland, the main theme was the ‘fourth industrial revolution’. Advances in robotics and artificial intelligence would have the transformative effect that steam power, electricity and ubiquitous computing achieved in previous centuries.
At Davos, the elite were told by Sebastian Thrun, the inventor of Google’s self-driving cars and an honorary professor at Delft University of Technology, that “almost every established industry is not moving fast enough” to adapt their businesses to this change. He suggested self-driving cars would make millions of taxi drivers redundant, while planes running solely on autopilot would remove the need for thousands of human pilots. However, don’t worry, as Thrun was optimistic that redundant roles will quickly be replaced: “With the advent of new technologies, we’ve always created new jobs.”
As one of the most prominent observers of the new ‘industrial revolution’, Erik Brynjolfsson, the Massachusetts Institute of Technology professor and co-author of The second machine age, put it, “We’re moving to a world where there will be vastly more wealth and vastly less work.” But he went on:
I think the biggest immediate change will be a move away from … one person [staying] in one profession or one job during their lifetime … That shouldn’t be a bad thing, and shame on us if we turn it into a bad thing.
Such optimism contrasted with the WEF’s own book launched at Davos, which reckoned that increased automation and AI in the workforce will lead to the loss of 7.1 million jobs over the next five years in 15 leading economies, while helping create just 2 million new jobs over the same period. In the financial sector, a thinking, learning and trading computer may well make even today’s superfast, ultra-complex investment algorithms (algos) look archaic - and possibly render human fund managers redundant. You might not care too much about the loss of hedge fund managers. But AI and robots will destroy the jobs of millions in productive sectors paying much less money. This is the prospect for labour in a robot-led capitalism.5
And it is not just the loss of jobs for millions that is the prospect arising from AI/robots, but some have argued that AI threatens the existence of humanity itself. Ray Kurzweil, the American inventor and futurist, has predicted that by 2045 the development of computing technologies will reach a point at which AI outstrips the ability of humans to comprehend and control it. Stephen Hawking has argued that “the development of full artificial intelligence could spell the end of the human race”.6 And Elon Musk, the founder of SpaceX and Tesla Motors, believes that AI is “potentially more dangerous than nukes”.7 The “biggest existential threat” to humanity, he thinks, is a Terminator-like super-machine-intelligence that will one day dominate humanity.
Moreover, while computers are quicker, smarter and shorn of human behavioural biases, they come with their own weaknesses. Disaster can strike quickly. For example, Knight Capital, a high-frequency trading firm, imploded in 2012 when its computers ran amok, in practice losing $10 million a minute in a devastating 45-minute trading blitz. As Gavekal, an investment brokerage, acerbically noted at the time,
Sometimes all computers do is replace human stupidity with machine stupidity. And, thanks to speed and pre-programmed conviction, machine stupidity can devour markets far faster than any human panic can achieve.
Algos based on artificial intelligence techniques may be the next generation of quantitative finance, but even industry insiders say they can unravel when confronted with the chaotic reality of markets: “This stuff in the hands of the wrong people can be very dangerous,” says Tom Doris, the head of Otas Technologies.8 For example, self-driving cars suffered twice as many accidents as human-driven ones in 2013, according to a University of Michigan study. 9 Most of them were minor scrapes, and the human drivers were at fault in every case, but this is a vivid illustration of how accidents can happen when man meets machine - whether on the road or in markets.
But can robots really replace humans within 30 years? Many doubt it. Scenarios such as Kurzweil’s are extrapolations from Moore’s law, according to which the number of transistors in computers doubles every two years, delivering greater and greater computational power at ever-lower cost.
But Gordon Moore, after whom Moore’s law is named, has himself acknowledged that his generalisation is becoming unreliable, because there is a physical limit to how many transistors you can squeeze into an integrated circuit. In any case, Moore’s law is a measure of computational power, not intelligence. A vacuum-cleaning robot, a Roomba, will clean the floor quickly and cheaply and increasingly well, but it will never book a holiday for itself with my credit card.
Luciano Floridi at the University of Oxford agrees that machines can do amazing things,10 often better than humans. For instance, IBM’s Deep Blue computer played and beat the former world champion, Garry Kasparov, at chess in 1997. In 2011, another IBM machine, Watson, won an episode of the TV quiz show Jeopardy, beating two human players, one of whom had enjoyed a 74-show winning streak. But Deep Blue and Watson are versions of the ‘Turing machine’, a mathematical model devised by Alan Turing, which sets the limits of what a computer can do. A Turing machine has no understanding, no consciousness, no intuitions - in short, nothing we would recognise as a mental life. It lacks the intelligence even of a mouse.
Floridi explains that in 1950 Turing proposed the following test. Imagine a human judge who asks written questions to two interlocutors in another room. One is a human being, the other a machine. If, for 70% of the time, the judge is unable to tell the difference between the machine’s output and the human’s, then the machine can be said to have passed the test. Turing thought that computers would have passed the test by the year 2000. He was wrong. Eric Schmidt, the former chief executive of Google, believes that the Turing test will be passed by 2018. So far there has been no progress. Computer programs still try to fool judges by using tricks developed in the 1960s.
For example, in the 2015 edition of the Loebner Prize, an annual Turing test competition, a judge asked: “The car could not fit in the parking space because it was too small. What was too small?” The software that won that year’s consolation prize answered: “I’m not a walking encyclopaedia, you know.”
Law of value
Are we entering a new industrial revolution like the early 19th century that will give capitalism a new lease of life in developing the productive forces, even if it means loss of jobs for hundreds of millions and rising inequality of income and wealth? Or are the new ‘disruptive technologies’ just a mirage that will change little in increasing economic growth and productivity, as Gordon argues? I think it is both, depending on the time and the cyclical eruptions that is the capitalist mode of production.11
Consider the impact of robots and AI seen through the prism of Marx’s law of value under capitalism. There are two key assumptions that Marx makes in order to explain the laws of motion under capitalism: (1) that only human labour creates value; and (2) over time investment by capitalists in technology and means of production will outstrip investment in human labour-power - to use Marx’s terminology, there will be a rise in the organic composition of capital.
Marx explained in detail in Capital that a rising organic composition of capital is one of the key features in capitalist accumulation. Investment under capitalism takes place for profit only, not to raise output or productivity as such. If profit cannot be sufficiently raised through more labour hours (ie, more workers and longer hours) or by intensifying efforts (speed and efficiency - time and motion), then the productivity of labour (more value per labour-hour) can only be increased by better technology. So, in Marxist terms, the organic composition of capital (the amount of machinery and plant relative to the number of workers) will rise in a secular fashion. Workers can fight to keep as much of the new value that they have created as part of their ‘compensation’, but capitalism will only invest for growth if that wage share does not rise so much that it causes profitability to decline. So capitalist accumulation implies a falling share going to labour over time, or what Marx would call a rising rate of exploitation (or surplus value).
The ‘capital-bias’ of technology is something continually ignored by mainstream economics. But as Branco Milanovic has pointed out, even mainstream economic theory could encompass this secular process under capitalist accumulation. As Milanovic puts it,
In Marx, the assumption is that more capital-intensive processes are always more productive. So capitalists just tend to pile more and more capital and replace labour ... This in Marxist framework means that there are fewer and fewer workers who obviously produce less (absolute) surplus value and this smaller surplus value over an increased mass of capital means that the rate of profit [r] goes down ...
The result is identical if we set this Marxist process in a neoclassical framework and assume that the elasticity of substitution is less than 1. Then, simply, r shoots down in every successive round of capital-intensive investments until it practically reaches zero. As Marx writes, every individual capitalist has an interest to invest in more capital-intensive processes in order to undersell other capitalists, but, when they all do that, the rate of profits decreases for all. They thus work ultimately to drive themselves ‘out of business’ (more exactly they drive themselves to a zero rate of profit).12
Milanovic then considers robot technology:
Net income, in Marxist equilibrium, will be low because only labour produces ‘new value’ and, since very few workers will be employed, ‘new value’ will be low (regardless of how high capitalists try to drive the rate of surplus value). To visualise Marxist equilibrium, imagine thousands of robots working in a big factory with only one worker checking them out, and with the useful life of robots being one year, so that you keep on replacing robots continuously and thus run enormous depreciation and reinvestment costs every year. The composition of GDP would be very interesting. If total GDP is 100, we could have consumption = 5, net investment = 5 and depreciation = 90. You would live in a country with GDP per capita of $500,000 - but $450,000 of that would be depreciation.
This poses the key contradiction of capitalist production: rising productivity leads to falling profitability, which periodically stops production and productivity growth. But what does this all mean, if we enter the extreme (science fiction?) future where robotic technology and AI lead to robots making robots, torobots extracting raw materials and making everything and carrying out all personal and public services, so that human labour is no longer required for any task of production at all?
Let us imagine a totally automated process where no human existed in the production. Surely, value has been added by the conversion of raw materials into goods without humans? Surely, that refutes Marx’s claim that only human labour can create value?
But this confuses the dual nature of value under capitalism: use-value and exchange-value. There is use-value (things and services that people need); and exchange-value (the value measured in labour-time and appropriated from human labour by the owners of capital and realised by sale on the market). In every commodity under the capitalist mode of production, there is both use-value and exchange-value. You cannot have one without the other under capitalism. But the latter rules the capitalist investment and production process, not the former.
Value (as defined) is specific to capitalism. Sure, living labour can create things and undertake services (use-value). But value is the substance of the capitalist mode of producing things. Capital (the owner) controls the means of production created by labour and will only put them to use in order to appropriate value created by labour. Capital does not create value itself.
But in our hypothetical, all-encompassing robot/AI world, productivity (of use-values) would tend to infinity, while profitability (surplus value to capital value) would tend to zero. Human labour would no longer be employed and exploited by capital (owners). Instead, robots would do all. This is no longer capitalism. It is more like a slave economy, as in ancient Rome.
In ancient Rome, over hundreds of years, the formerly predominantly small-holding peasant economy was replaced by slaves in mining, farming and all sorts of other tasks. This happened because the booty of the successful wars that the Roman republic and empire conducted included a mass supply of slave labour. The cost to the owners of these slaves was incredibly cheap (to begin with), compared with employing free labour. The slave-owners drove the farmers off their land through a combination of debt demands, requisition in wars and sheer violence. The former peasants and their families were forced into slavery themselves or into the cities, where they scraped a living with menial tasks and skills or begged. The class struggle did not end. The struggle was between the slave-owning aristocrats and the slaves and between the aristocrats and the atomised plebs in the cities.
In the completely automated planet, how would the goods and services produced by robots be distributed in order to be consumed? That would depend on who owns the robots, the means of production. Suppose there are 100 lucky guys on the robot-run planet. One of them may own the best robots and so appropriate the whole product. Why should he share it with the other 99? They will be sent back to the Earth. Or they might not like it and will fight for the appropriation of some of the robots. And so, as Marx put it once, the whole shit begins again, but with a difference.
All will depend on how humanity would get to a completely automated society. On the basis of a socialist revolution and common ownership, the distribution of the output produced by the robots can be controlled and distributed to each according to his/her needs. If society operates on the basis of a continuation of the private ownership of the robots, then the class struggle for the control of the surplus continues.
The question often posed at this point is: who are the owners of the robots and their products and services going to sell to make a profit? If workers are not working and receiving no income, then surely there is massive overproduction and underconsumption? So, in the last analysis, it is the underconsumption of the masses that brings capitalism down?
Again, I think this is a misunderstanding. Such a robot economy is not capitalist any more; it is more like a slave economy. The owners of the means of production (robots) now have a super-abundant economy of things and services at zero cost (robots making robots making robots). The owners can just consume. They do not need to make ‘a profit’, just as the aristocrat slave-owners in Rome just consumed and did not run businesses to make a profit. This does not deliver an overproduction crisis in the capitalist sense (relative to profit) nor ‘underconsumption’ (lack of purchasing power or effective demand for goods on a market), except in the physical sense of poverty.
Mainstream economics continues to see the rise of the robots under capitalism as creating a crisis of underconsumption. As Jeffrey Sachs put it, “Where I see the problem on a generalised level for society as a whole is if humans are made redundant on an industrial scale (47% quoted in US), then where’s the market for the goods?”13Or as Martin Ford writes,
there is no way to envision how the private sector can solve this problem. There is simply no real alternative except for the government to provide some type of income mechanism for consumers.14
Ford does not propose socialism, of course, but merely a mechanism to redirect lost wages back to ‘consumers’, but such a scheme would threaten private property and profit.
A robotic economy could mean a super-abundant world for all (post-capitalism, as Paul Mason suggests15); or it could mean Elysium. FT columnist Martin Wolf put it this way:
The rise of intelligent machines is a moment in history. It will change many things, including our economy. But their potential is clear: they will make it possible for human beings to live far better lives. Whether they end up doing so depends on how the gains are produced and distributed. It is possible that the ultimate result will be a tiny minority of huge winners and a vast number of losers. But such an outcome would be a choice, not a destiny. A form of techno-feudalism is unnecessary. Above all, technology itself does not dictate the outcomes. Economic and political institutions do. If the ones we have do not give the results we want, we must change them.16
It is a social ‘choice’ or, more accurately, it depends on the outcome of the class struggle under capitalism. John Lanchester is much more to the point:
It’s also worth noting what isn’t being said about this robotified future. The scenario we’re given - the one being made to feel inevitable - is of a hyper-capitalist dystopia. There’s capital, doing better than ever; the robots, doing all the work; and the great mass of humanity, doing not much, but having fun playing with its gadgets …
There is a possible alternative, however, in which ownership and control of robots is disconnected from capital in its current form. The robots liberate most of humanity from work, and everybody benefits from the proceeds: we don’t have to work in factories or go down mines or clean toilets or drive long-distance lorries, but we can choreograph and weave and garden and tell stories and invent things and set about creating a new universe of wants. This would be the world of unlimited wants described by economics, but with a distinction between the wants satisfied by humans and the work done by our machines.
It seems to me that the only way that world would work is with alternative forms of ownership. The reason, the only reason, for thinking this better world is possible is that the dystopian future of capitalism-plus-robots may prove just too grim to be politically viable. This alternative future would be the kind of world dreamed of by William Morris, full of humans engaged in meaningful and sanely remunerated labour. Except with added robots.
It says a lot about the current moment that, as we stand facing a future which might resemble either a hyper-capitalist dystopia or a socialist paradise, the second option doesn’t get a mention.17
In the meantime, capitalism is grappling with these new and ‘disruptive’ technologies. And there is still little sign of any significant return to the previous trend in business investment growth. In 2013, real spending on business investment in the US rose 3.8% - little more than half the rate achieved prior to great recession. And what is especially noticeable is that spending on hi-tech innovatory equipment - the previously dynamic, high-growth sector, with an average of 10%-20% annual growth - is very weak, now growing at a pace slower than overall real GDP.
Hi-tech spending on both equipment and software has fallen as a share from 4.7% of US GDP in 2000 to 3.5% in 2013. It is this area that is key to boosting productivity. What is the reason for this slowdown in investment in new technology? Well, it appears to be that the cost of new equipment and software is just too high relative to the realised and expected return on those investments - in other words, the rate of profit is not high enough. Indeed, as I have argued at length before, the major capitalist economies are still locked into what I call a long depression - of below-trend real GDP and productivity growth and a debt deflationary environment.
But if the end of this long depression does not lead to the replacement of the capitalist mode of production through political action from energised working class movements, capitalism will recover - as Marx said, there is no permanent crisis.
Yes, crises are endogenous to capitalism, because of the main contradiction within the capitalist mode of production: accumulation for profit and not need. But also it is possible for capitalism to recover and soldier on ‘endogenously’, when sufficient old capital is destroyed in value (and sometimes physically) to allow for a new period of rising profitability. Capitalism can only be replaced by a new system of social organisation through the conscious action of human beings - in particular by the majority class of people, the working class globally. Without such conscious action, capitalism can stumble on.
In the third section of Gordon’s book, he looks at why productivity growth did soar at one particularly notable juncture in the 1930s. Gordon reckons that the great depression was a period of innovation that “directly contributed to the great leap” in the 1940s. Gordon also points to the “high-pressure learning-by-doing that occurred during the high-pressure economy of World War II.” World War II gave America its first jet aircraft (the Bell P-59), mass-produced penicillin and nuclear power. Perhaps even more important, factories like Henry Kaiser’s shipyards taught managers and workers how to radically speed up production. Something similar could happen when this long depression ends, as it will.
Also, capitalism could get a further kick forward from exploiting the hundreds of millions coming into the labour forces of Asia, South America and the Middle East. This would be a classic way of compensating for the falling rate of profit in the mature capitalist economies. While, as unproductive labour has risen sharply, the industrial workforce in the mature capitalist economies has shrunk to under 150 million: in the so-called emerging economies the industrial workforce now stands at 500 million, having surpassed the industrial workforce in the imperialist countries by the early 1980s. In addition, there is a large reserve army of labour composed of unemployed, underemployed or inactive adults of another 2.3 billion people globally, who could also be exploited for new value.18
So there may be life in global capitalism yet, even if it is in ‘down mode’ right now. Or maybe this potential labour force will not be ‘properly exploited’ by the capitalist mode of production and Gordon is right. The world rate of profit (not just the rate of profit in the mature G7 economies) stopped rising in the late 1990s and has not recovered to the level of the golden age for capitalism in the 1960s, despite the massive potential global labour force.19
It seems that the countervailing factors of foreign investment in the emerging world, combined with new technology, have not been sufficient to push up the world rate of profit in the last decade or so, so far. The downward phase of the global capitalist cycle is still in play.
1 . RJ Gordon The rise and fall of American growth Princeton 2016.
2 . www.ft.com/cms/s/2/7d9874c0-a25d-11e5-8d70-42b68cfae6e4.html#axzz40QrlsiMH.
3 . ‘Malthus, Marx and modern growth’: www.project-syndicate.org/commentary/kenneth-rogoff-identifies-several-obstacles-to-keeping-living-standards-on-an-upward-trajectory?barrier=true.
4 . www.voxeu.org/article/technological-progress-thing-past.
5 . See https://thenextrecession.wordpress.com/2015/08/23/robots-and-ai-utopia-or-dystopia-part-one.
6 . www.bbc.co.uk/news/technology-30290540.
7 . www.cnbc.com/2014/08/04/ai-potentially-more-dangerous-than-nukes-musk-warns.html.
8 . www.ft.com/cms/s/0/5eb91614-bee5-11e5-846f-79b0e3d20eaf.html#axzz40QrlsiMH.
9 . www.umich.edu/~umtriswt/PDF/UMTRI-2015-34_Abstract_English.pdf.
10 . www.dailymail.co.uk/sciencetech/article-3417826/No-AI-version-Godzilla-enslave-Oxford-professor-says-intelligent-robots-HELP-humanity.html.
11 . See https://thenextrecession.wordpress.com/2015/08/29/robots-and-ai-utopia-or-dystopia-part-two.
12 . http://glineq.blogspot.co.uk/2015/04/the-rule-of-robots-in-stiglitz-and-marx.html.
13 . http://prospect.org/article/how-live-happily-robots.
14 . www.npr.org/sections/alltechconsidered/2015/05/18/407648886/attention-white-collar-workers-the-robots-are-coming-for-your-jobs.
15 . See https://thenextrecession.wordpress.com/2015/07/21/paul-mason-and-postcapitalism-utopian-or-scientific.
16 . Financial Times February 4 2014.
17 . www.lrb.co.uk/v37/n05/john-lanchester/the-robots-are-coming.
18 . See https://thenextrecession.wordpress.com/2014/10/21/de-industrialisation-and-socialism.
19 . See https://thenextrecession.wordpress.com/2014/04/23/a-world-rate-of-profit-revisited-with-maito-and-piketty.