This is the third and final post on the issue of robots and artificial intelligence (AI). In the first post, I argued that while robots and AI are a leap forward in mechanisation and automation, they will not do away with the basic contradiction within the capitalist mode of production between the drive to raise the productivity of labour and the profitability of capital over time. As I said “over time, a capital-bias or labour shedding means less new value is created (as labour is the only form of value) relative to the cost of invested capital. There is a tendency for profitability to fall as productivity rises. In turn, that leads eventually to a crisis in production that halts or even reverses the gain in production from the new technology. This is solely because investment and production depend on the profitability of capital in our modern mode of production.”
In the second post, I considered in more detail how the law of value that dominates the profit-making capitalist mode of production would be affected by the hypothetical (or real?) possibility of a fully automated economy where no human labour is expended at all. “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.”
But I argued that before this state of ‘singularity’ (as it is called) was reached, capitalism as system would have broken down. “We would never get to a robotic society; we would never get to a workless society – not under capitalism. Crises and social explosions would intervene well before that… accumulation under capitalism would cease well before robots took over fully, because profitability would disappear under the weight of ‘capital-bias’.”
In this third post, I want to consider just how likely it is that highly intelligent robots will take over the world of work (and maybe the world) in the near future. It’s my contention that, despite the optimism of the AI and robot drivers, it’s not going to happen soon.
What is true is that the use of robots is rising fast. The level of robotics use has almost always doubled in the top capitalist economies in the last decade. Japan and Korea have the most robots per manufacturing employee, over 300 per 10,000 employees, with Germany following at over 250 per 10,000 employees. The United States has less than half the robots per 10,000 employees compared to Japan and The Republic of Korea. The adoption rate of robots increased in this period by 40% in Brazil, by 210% in China, by 11% in Germany, by 57% in The Republic of Korea, and by 41% in the United States.
This development has been called a ‘second wave of automation’, one that is centered on artificial cognition, cheap sensors, machine learning and distributed smarts. This deep automation will touch all jobs, from manual labor to knowledge work. And it is reducing employment, just as mechanisation under previous industrial revolutions did.
Andrew McAfee, the coauthor with his MIT colleague Erik Brynjolfsson of The Second Machine Age, has been one of the most prominent figures describing the possibility of a “sci-fi economy” in which the proliferation of smart machines eliminates the need for many jobs. (see “Open Letter on the Digital Economy,” in which McAfee, Brynjolfsson, and others propose a new approach to adapting to technological changes.) Such a transformation would bring immense social and economic benefits, he says, but it could also mean a “labor-light” economy.
Hod Lipson says “More and more computer-guided automation is creeping into everything from manufacturing to decision making”. Prominent Columbia University economist Jeffrey Sachs recently predicted that robots and automation would soon take over at Starbucks. But there are good reasons to believe that Sachs and others could be wrong. The success of Starbucks has never been about getting coffee more cheaply or efficiently. Consumers often prefer people and the services humans provide. Take the hugely popular Apple stores, says Tim O’Reilly, the founder of O’Reilly Media. Staffed by countless swarming employees armed with iPads and iPhones, the stores provide a compelling alternative to a future of robo-retail; they suggest that automating services is not necessarily the endgame of today’s technology. “It’s really true that technology will take away a class of jobs,” says O’Reilly. “But there is a choice in how we use technology.”
And just how close are AI robots to doing all human work? AI researchers have noted that the simplest tasks for humans, such as reaching into a pocket to retrieve a quarter, are the most challenging for machines. For example, iRobot’s Roomba robot is autonomous, but the vacuuming task it performs by wandering around rooms is extremely simple. By contrast, the company’s Packbot is more expensive, designed for defusing bombs, but must be teleoperated or controlled wirelessly by people.
The Defense Advanced Research Projects Agency, a Pentagon research arm, held a Robotics Challenge competition in Pomona, Calif. Theer was $2 million in prize money for the robot that performs best in a series of rescue-oriented tasks in under an hour. In the previous contest in Florida in December 2013, the robots, which were protected from falling by tethers, were glacially slow in accomplishing tasks such as opening doors and entering rooms, clearing debris, climbing ladders and driving through an obstacle course. (The robots had to be placed in the vehicles by human minders.) Reporters who covered the event resorted to such analogies as “watching paint dry” and “watching grass grow.”
This time, the robots had an hour to complete a set of eight tasks that would probably take a human less than 10 minutes. And the robots failed at many. Most of their robots were two-legged, but many had four legs, or wheels, or both. But none were autonomous. Human operators guided the machines via wireless networks and were largely helpless without human supervisors. Little headway has been made in “cognition,” the higher-level humanlike processes required for robot planning and true autonomy. As a result, any researchers have begun to think instead of creating ensembles of humans and robots, an approach they describe as co-robots or “cloud robotics.”
So there’s still a long way to go. David Graeber has also raised other obstacles to the fast adoption of autonomous AI fully automated robots, namely the capitalist system itself. Funding for new technology does no go into solving the needs of people and reducing human toil as such, but into what will raise profitability. “Once upon a time, he said, “when people imagined the future, they imagined flying cars, teleportation devices and robots who would free them from the need to work. But strangely, none of these things came to pass.”
What happened, instead, was that industrialists poured research funds not into the invention of the robot factories that everyone was anticipating in the sixties, but into relocating their factories to labor-intensive, low-tech facilities in China or the Global. And governments shifted funds into military research, to weapons projects, research in communications and surveillance technologies and similar security-related concerns. “One reason we don’t have robot factories yet is because roughly 95 percent of robotics research funding has been channeled through the Pentagon, which is more interested in developing unmanned drones than in automating paper mills.”
William Nordhaus from Yale University’s department of economics, has tried to estimate the future economic impact of AI and robots (SSRN-id2658259). Nordhaus reckons ‘singularity’and its impact is still a long way away. Consumers may love their iPhones, but they cannot eat the electronic output. Similarly, at least with today’s technologies, production requires scarce inputs (“stuff”) in the form of labour, energy, and natural resources, as well as information for most goods and services. Nordhaus says projecting the trends of the last decade or more, it would be in the order of a century before growth variables would reach the level associated with a growth-focused singularity.
Nordhaus also raises the issue of robots out of control – robots rule the world including us. “The development of superintelligence raises a new concern not contemplated before in the development of political and military spying and weapons. We must be concerned that to the list of adversaries will be added the superintelligent machines themselves…. Will the superintelligent treat us as flies to wonton boys?” So there’s one job category for humans that won’t be easily eliminated: defending our interests from the all-encompassing power of the AI: “We routinely spend 5% of output on defense, and this might rise to a much larger number when faced with a more powerful enemies like superintelligent machines. So one occupation at least would survive into the Era of Singularity.”
But let’s not throw the baby out with the bathwater. Technical advances to meet the needs of people, to help end poverty and create a society if superabundance without damaging the environment and the ecology of the planet is what we want. If AI/robotic technology can bring us closer to that, all the better.
But the obstacle to a harmonious superabundant society based on robots reducing human toil to a minimum is Capital. In its Future of Work report last year, UKCES came up with a number of scenarios which included both the possibility of a long period of stagnation and of a technology-driven productivity leap. One thing all the scenarios had in common, though, was that, for those without good skills, powerful connections or inherited wealth, the future looks extremely bleak. The Economist concluded, at the end of a long piece on technology and work last year: “S]ociety may find itself sorely tested if, as seems possible, growth and innovation deliver handsome gains to the skilled, while the rest cling to dwindling employment opportunities at stagnant wages.” Or, as John Naughton put it, “a concierge economy [with] legions of network co-ordinated serfs.”
While the means of production (and that will include robots) are owned by a few, the benefits of a robot society will accrue to the few. Whoever owns the capital will benefit as robots and AI inevitably replace many jobs. If the rewards of new technologies go largely to the very richest, as has been the trend in recent decades, then dystopian visions could become reality. I quote again from John Lanchester: “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.”
Let me sum up the conclusions of my posts on robots and AI.
- The new technology of robots and AI is coming fast. As in all technology under capitalism, it has a ‘capital-bias’; it will replace human labour. But under capitalism, that capital bias is applied to reduce costs and boost profitability not meet people’s needs.
- Robots and AI will intensify the contradiction under capitalism between the drive by capitalists to raise the productivity of labour through ‘mechanisation’ (robots) and the resulting tendency for the profitability in this investment for the owners of capital to fall. This is Marx’s most important law in political economy – and it becomes even more relevant in the world of robots. Indeed, the biggest obstacle to a world of super-abundance is capital itself. Well before we get to ‘singularity’ (if we ever do) and human labour is totally replaced, capitalism will experience an increasingly deeper series of man-made economic crises.
- Robot technology will reduce many existing jobs (and create some new jobs) and is doing so already. But singularity and a robot world is still a long way away. That is because the AI technology is not being directed by capital into the most productive areas but into the most profitable (not the same thing). And the costs of ‘controlling’ AI robots will increase.
- A super-abundant society where human toil is reduced to a minimum and poverty is eliminated won’t happen unless the ownership of the means of production changes from private control (capitalist oligarchy) to ownership in common (democratic socialism). That’s the choice between utopia and dystopia.
16 thoughts on “Robots and AI: utopia or dystopia? – part three”
Pretty much agree with the whole series. A nit pick, Nordhaus doesn’t seem to be taking exponential development in to account. Moore’s Law is still in effect.
“A super-abundant society … won’t happen unless the ownership of the means of production changes from private control (capitalist oligarchy) to ownership in common (democratic socialism). That’s the choice between utopia and dystopia.”
Agreed, except for the fact that “super-abundant” sounds a bit to materialistic or even hedonistic. The question is if and how this can come to happen. As long as the (far) left and most of the Marxists remain in the frame of planned economy (doing away with the market and organizations as independent entities with their own interests), class struggle leading to a revolutionary overthrow of the capitalist system (in fact meaning the elimination of the upper class, not the transformation) and the abolishment of any stratification I don’t see much hope.
On the side: many human activities that are toil under capitalist relations are in fact satisfying and fulfilling activities that shouldn’t be replaced by robot technology. More on http://onrust.2fd.eu/2014/05/capital-should-be-in-solid-hands/
How much can the ongoing decline in productivity growth rates and productive investment be sketched out in relation to this? I assume they’d tend in the long-term to decline alongside the falling rate of profit, as capital increasingly diverts into an “unproductive” parasitic layer, speculating or competing in the distributional arena of mass consumerism, resulting in relative stagnation. In that Baffler article, Graeber points out that the increased substitution of cheap mass labor for improvements in factory automation, and the general failure of the information “revolution” to make good on its very real potential, would make a lot of sense if Marx’s analysis were true. But of course he rejects that possibility for the same reason he seems convinced that “bullshit jobs” are purely a political or cultural problem, a problem of arbitrary “bureaucracy” holding down creativity, a circumstance he thinks capitalism has almost randomly found itself mired in. Not by any systematic contradiction within capitalism.
The reason for chaining people to labour for survival is not primarily to maximise profits, although the logic of the system demands it. The capitalists realised early on that a leisure or ‘playful work’ society is ideologically infertile ground for capitalism. And so is abundance; capitalism is wedded to scarcity in some form or other. For that reason alone, regardless of the pure economics of it, the imagined utopia of, say a 12-15 hours work week is impossible under capitalism.
We should also factor in the dimension of hierarchy. Capitalists don’t want to rule over machines, that doesn’t impress potential mates!
They want control over other human beings, this way they have status. This way they impress potential mates.
While ever you have capitalism or class society forget about robots doing all the work.
Actually forget about robots doing all the work, it is will never happen, is pure fantasy.
Automation reduces the amount of socially necessary labour time required to produce goods and services. In the words of contemporary “economics”, automation is the main driver for increasing “productivity”.
At the moment, in Australia, there are 750,000 people chasing after 150,000 job openings. As automation accelerates, this ratio will only get worse, if the official, legislated 40 hour work week remains constant. Unemployment will grow unless the workers begin to understand that the official work week needs to shortened due to productivity increases. This understanding will have to be coupled with political and industrial organising, just as it was during the class struggle for the 8-hour day. Otherwise, workers will be reduced to playing the role of “surplus population” and become a mass of broken wretches.
” …the workers begin to understand”? … “political and Industrial organising”?? Good luck with that Mike; particularly when Australia’s alternative ‘left-wing’ Prime Minister-in-waiting is intimately connected to key individuals/families and institutions of the Right.
Moreover, the remaining 11 million or so Australians still clinging to their only source of livelihood are not yet ready to ‘down tools’ and storm the parliament … particularly when employers remain free to import ‘cheap labour’ at will via the (ab)use of 457 visas and shonky ‘Free Trade’ agreements!
A. I. is to computing what perpetual motion was to 17th century mechanics. Hopeless idealism and hyperbole. Far from computing rising to the level of intelligence, intelligence is decsending to computing. As always with capitalism they take the creative worker, turn them first into a cog and finally into a mere machine monitor.
I wonder on what actual experience and knowledge your valuation of AI and HI is grounded. My readings on both fields so far point quite another way.
“I wonder on what actual experience and knowledge your valuation of AI and HI is grounded”
I have played Max Payne 3 and I have to agree with David Ellis!
Seriously though, in my experience computers tend to replace people, systems are introduced to cut costs. You might think that bringing in computers would free up the time associated with simple calculations and allow people to be creative. Or turn data in relevant information. But under capitalism this process is bastardized, perverted, haphazard etc etc etc.
While capitalism has an historically impressive (and destructive) dynamism it doesn’t handle that dynamism too well at all, which I think was a central plank of Marxist criticism.
It’s worth noting that this optimism is congenial with the academic research of cognitive science. From what I’ve seen, the mainstream view is a kind of computationalism/functionalism (also biological determinism): the consciousness is merely a function of the brain, and the brain is a biological information processing machine, therefore other information processing machines (like computers) can become conscious too.
What’s curious is that this is completely antithetical to 3 of the most influential philosophers of our age: Heidegger, later era Wittgenstein, and of course Marx. The cognitivists still cling to a kind of residual Cartesianism, only their homonculus is the brain itself. Neuroscientists inherited an obsolete philosophy.
As Marxists say, consciousness is produced by the dialectical interplay of the body and its environment, and since humans produce and reproduce their environments, this also means that consciousness (the prerequisite of understanding) is the product of the society the individual lives in. Which leads us to the paradox that the only way machines can become sapient, if they participate in human society, which in capitalism includes being a wage laborer and a consumer, which is exactly contrary to the reason capitalists are funding AI developers.
“Which leads us to the paradox that the only way machines can become sapient, if they participate in human society, which in capitalism includes being a wage laborer and a consumer.”
There is no reason at all to believe that AI has to participate in this prescribed way under capitalism. There are many other ways and these are partly already in use.
I used the word “sapient” for its connotation of Homo Sapiens, in other words I was referring to the research of “strong AI”. Certainly phones are “smarter” now than they were 20 years ago… and of course factories are better and faster at making cars than they were a 100 years ago and they are getting more efficient still. Automation is developing. But the strong AI program has mistaken presuppositions… and we’re not gonna have a fully robotic production without sentient machines, and we’re not gonna have sentient machines if they don’t have their own practical life activity.
Dear Michael, my name is Raquel Varela, I am a marxist labour historian, I am working with IDC on automation and restructuring of labour force, there is a email where I can send you a private email please? Thank you in advance, best wishes, Raquel
It could be also a political utopia: the sort of “computer-aided anarchy” imagined by Iain M. Banks (the Culture).
See Yannick Rumpala, “Artificial intelligences and political organization: an exploration based on the science fiction work of Iain M. Banks”, Technology in Society, Volume 34, Issue 1, 2012, available at: https://www.researchgate.net/publication/256987370_Artificial_intelligences_and_political_organization_An_exploration_based_on_the_science_fiction_work_of_Iain_M_Banks
Human jobs, or those you cite as still challenging to automatons coming as a result of choice and expediency would predominate lifetime.
Just like how we need not build mountains in order to receive relief rainfall or stitch ourselves onto the earth’s surface in order not to fall(gravity ensures it), Humans shall always use what they find as premises and construct new and varied passages of TIME