Peter Ross argues that the mass implementation of AI technology by capitalists necessitates a solid grasp of Marx’s Value Theory.
It is becoming clear that the revolution in artificial intelligence is a vital issue for the working class. Capitalists are betting big that AI will allow them to dramatically cut costs by firing millions of workers. Tens of billions of dollars are being frantically poured into the implementation of these technologies in practically every sector of the economy.
This process is already radicalizing workers and intensifying the class struggle. Striking writers and actors in the WGA and SAG-AFTRA have consistently described AI as an “existential threat” to their jobs. Millions more will soon come to similar conclusions.
What is the worth of human labor in a world with artificial neural networks? To the capitalist, profit appears to derive from the growth of production and the development of technology in general. If more can be produced on a given investment than the initial value of that investment, then a profit is made: revenue minus expenses, pure and simple. The growth of the productive forces and the process of capital accumulation are taken to be identical. At first glance, Marx’s contention that profit comes from human labor alone appears to be absurd. As machines grow more powerful and more subtle, and now even appear to possess a spark of intelligence, capitalist society becomes more confident that these must be a far greater source of wealth and profit. Soon, the capitalists and their ideologists imagine, human labor will not be needed at all and a fully-automated capitalism can surge ahead.
But for those seeking an alternative to capitalism, this cannot be accepted as a settled matter. Against the triumphalist claims of capital that the masses of workers can be replaced by machines, it is necessary for us to clarify the role of human beings in capitalist production and the real origins of profit.
Are “Causal Powers” the Origin of Value?
One recent attempt to grapple with the Marxian view on this question comes in Ian Wright’s article “Why Machines Don’t Create Value,” published in October 2021 in Cosmonaut, which prompted a response letter by Gil Schaeffer. Wright has previously done some interesting work in the promising field of econophysics1, but his Cosmonaut piece badly misrepresents Marx’s theory, and the response letter (which claims that value theory is “fatally flawed”) does no better. It is necessary to rebut Wright because, as we will see, he presents in Marx’s name a theory that rests on the same mystifications that underlie the bourgeois critiques of Marx. I do not claim to be any authority on economics, but have written what follows in an attempt to clarify some of the questions to myself, and in the hope that my own efforts can be useful to other Marxists as they do the same.
Wright begins by summarizing some basic concepts from Marx, but goes on to make the very strange claim that “The cause of profit, according to Marx, is human labor because it, and only it, can work harder and work smarter.” (I would be very curious to know where Marx makes this claim.) Wright then raises some apparent problems with this version of value theory. He notes that “any economic output is jointly caused by human work combined with non-human work” and that humans themselves may be thought of as machines (albeit of a very sophisticated kind). Hence, it seems impossible to draw a strict line between human labor and non-human labor in the complex process of capitalist production. Wright repeats the criticism of bourgeois economists that Marx simply defines labor as the source of value without ever supplying adequate proof: “Also, critics of Marx’s theory have, quite correctly, pointed out that the objective cost structure of an economy can be measured in multiple ways, not just by labor time. We can equally talk of surplus oil value, surplus corn value, or surplus energy value. In fact, any commodity that’s a basic input.”
Wright spends the rest of his essay trying to find a way out of this paradox. In order to defend the idea that labor is the only source of value, Wright thinks he needs to find a fundamental distinction between humans and machines, “A Turing Test for Marx’s theory of surplus-value.” His solution he calls the “causal powers thesis”: The distinction between human labor and machines, he says, is that “humans are endlessly inventive and creative and adaptable — our imaginations are prodigious, and we learn by doing. Animals, machines and plants simply don’t have these causal powers. Our causal powers are precisely those that can’t really be kept in a box, but will always overspill it.” Hence, for Wright, only human creativity can act to increase economic output: “So machines cannot in general act to change the level of profits. But humans can. And that’s what Marx’s theory of surplus-value is getting at.”
First of all, this is not an accurate representation of Marx, and it is worth noting that Wright doesn’t provide any textual evidence of Marx making this argument himself. Marx may refer to the unique creative capacities of human beings, but he does not do so as an explanation of value, which is a specific economic concept. In fact, Marx opens his Critique of the Gotha Programme by refuting the very notion that Wright is putting forward, that labor possesses a “supernatural creative power.” As Marx explains, “Labor is not the source of all wealth. Nature is just as much the source of use values (and it is surely of such that material wealth consists!) as labor, which itself is only the manifestation of a force of nature, human labor-power.” By the same token, labor is not the only source of an increase in wealth or economic output, which may also come from improved technique, an increased use of already-invented machines, or a heightened exploitation of nature.
While Wright uses the terminology of a change in profits, he implicitly uses this to mean an increase in productivity or total output. In fact, he sometimes switches to this terminology, for instance when he says that machines “act in a constant manner and lack the general capability to notice ways to get more outputs for fewer inputs.” [emphasis added] But profit and productivity are two different things. An increase in the productivity of a section of capital could be accompanied by a decrease in profit — for instance, if it coincides with an increase in wages — and vice versa. The whole question is what determines the portion of the social product that is appropriated as profit. There is no way to ascertain this through a treatment of production as a lump sum — as though each capitalist were trying to get more output for fewer inputs independently of the rest of the market — but only by considering the whole circuit of capital. For Marx, value is a social relation: it does not stem from the creative powers of humans but from the process of commodity exchange.
The development of the productive forces through technical innovation is mostly due to the application of science outside the productive process itself, not by the worker. Marx’s view is not that human labor in general is the source of profit. It is that wage labor is the source of profit under capitalism.
Let’s consider an example. Einstein invented the theory of special relativity during his spare time while working at a patent office, and his theory went on to revolutionize the productive forces. By Wright’s conception, Einstein was a tremendous source of profits, but by Marx’s conception, it was only his work at the patent office as a wage laborer that contributed to profits. This may at first seem like a shocking shortcoming of Marx’s theory, until we realize that he is not interested in tracking the total quantity of use values that can be produced over time but rather he is trying to explain what social forces drive the historical development of capitalism as a mode of production. The motive force of capitalism is not the production of wealth in general but of private profit, not the production of use values (the worth of a commodity to a user) but of exchange values (how much it is worth on the market in terms of other commodities).
Furthermore, Wright’s invocation of “agency” and “causal powers” as the origin of value is inconsistent and idealist. As Wright himself notes, Marx refers to the human worker in The Grundrisse as merely a cog in the social process of production, completely lacking in agency and subordinated to the machine operations of industry. Conversely, the “invisible hand of the market,” the “alien force” of capitalism, has its own form of agency and can be said to exert “causal powers” in its own right. Or as Wright quotes from Marx, “the worker’s activity [is] reduced to a mere abstraction of activity, is determined and regulated on all sides by the movement of machinery,” while on the other hand, “it is the machine which possesses skill and strength in place of the worker, is itself the virtuoso, with a soul of its own in the mechanical laws acting through it.”
Human labor does not possess any supernatural ability to act as the prime mover or ultimate cause of events, as Wright’s “causal powers thesis” suggests. Human labor-power, to repeat the above quote of Marx, is itself “only the manifestation of a force of nature.” Hence, the term “causal powers” is necessarily vague, as it expresses nothing more than the mysteries of cognition that we don’t understand.
The fundamental problem with Wright’s analysis is that he follows the bourgeois critics in interpreting value as a metaphysical category. Having done so, he has no choice but to resort to a defense that is itself based on metaphysical abstractions such as “agency” and “causal powers,” from which he tries to derive a labor theory of value. This is an idealist approach. In contrast, Marx, as he writes in his “Notes on Adolph Wagner’s Textbook of Political Economy,” starts from commodity exchange, and attempts to derive a theory that describes the workings of the capitalist economy. The abstraction of surplus value arises from this analysis, but the analysis starts from the material facts of social life.
If the content of Marx’s theory is that only human beings have creative powers, we would have to concede to the bourgeois critics that it is a theory based on a mysticism, and that even if it were somehow an accurate description of capitalism in Marx’s day, it is outmoded today and becoming more outmoded all the time. Wright says as much himself. Having committed himself to the claim that human labor is special because only humans have “causal powers,” Wright concludes that advanced AI will eventually also be able to create value. But today’s AI models already exhibit a certain degree of intelligence, creativity, and agency. Would Wright argue that ChatGPT is a source of value?
Furthermore, Wright’s theory is ahistorical and applies equally to any human society. This robs value theory of its whole purpose, which is to explain the historical evolution of capitalism and show how its own development leads to its ultimate dissolution. Marx’s is a theory only of capitalism. As he makes clear, a slave, whatever causal powers they may possess, produces no surplus value.
Why Machines (and Slaves) Don’t Produce Surplus Value
Under capitalism, the key difference between wage labor and machinery is that machines and their products are bought and sold as commodities — they are direct inputs and outputs of the market — whereas wage laborers buy commodities on the market and sell a commodity in the form of their labor-power.
Let us assume for the moment that we are dealing with a competitive market, in which machinery and labor-power are equally accessible to all sections of capital. In this case, it is clear that a capitalist cannot derive profit from machinery, any more than from any other commodity produced on the market. To see this, suppose that a robot somehow produces a quantity of commodities over its lifetime that is worth more on the market than the robot’s production cost. Then, in a competitive market, investment will flow into the production of this robot and more of the commodity will be produced. The price of the commodity will then fall until it reaches the price of production of the robot. At this point, no profit can be derived from the robot because the cost of purchasing the robot is equal to the cost of the goods it produces over its lifetime. It is for this reason that machines can be treated as constant capital, which pass all of their value into their products. 2
The dictatorship of the capitalist class over the means of production is what enables it to capture everything that is produced using these means of production minus the wages of the workers employed (which, under perfect market competition, are driven to subsistence levels). This class monopoly is the source of the capitalists’ ability to extract a profit, the reason labor-power is the only commodity “whose use-value possesses the peculiar property of being a source of value.” (Capital, Vol. 1) Out of the collection of commodities produced in a given workday, capital needs only to sell a portion back to workers. The difference between the two is surplus value. As Marx expresses, “The capitalist’s profit is derived from the fact that he has something to sell for which he has paid nothing.” (Capital, Vol. 3, Ch. 2) The ability to extract something for nothing can be based only on unequal power — that since the worker has nothing to sell but their labor-power, they are forced to participate in the market on unequal terms.
We can push our robot thought experiment a step further by supposing that the robot can reproduce itself. In this case, we are no longer dealing with capitalism — because the robot is not the output of a social division of labor — but rather with slavery. Suppose that the robot/slave works 10 hours a day, covering its own maintenance and reproduction in less time, say 4 hours, and producing a widget in the remaining 6 hours. Clearly, under perfect market competition, the cost of this widget on the market will fall to 0 (since the robot’s price of production is 0). The widget may have a use value, but it has no exchange value: the owner can realize a surplus product of widgets, but will be unable to accrue a profit from exchanging them on the market. For this reason, the robot/slave cannot be considered to be a source of surplus value. (This discussion was motivated by the comments by Arthur Bough (Boffy) on Michael Roberts’ 2018 article “Robots: what do they mean for jobs and incomes?”)
Of course, the assumption of perfect competition is not realistic, but removing this assumption does not change anything essential. Suppose that a certain capitalist has a monopoly on a robot that produces a certain commodity. In this case, the monopolist can charge an amount equal to whatever it would cost to produce the same commodity without the use of their robot, and pocket the difference as an additional profit. But they can realize their profits only to the extent that their commodities can be realized as exchange values on the rest of the market. The costs that they save are equal to the costs that the non-monopolists lose, so the monopolist has done no more than parasitize profit from their competitors.
So machines, no matter how advanced, cannot be the source of profit, not because they lack “causal powers,” but because they are themselves commodities in the capitalist market.
Once again, the above is only an attempt to sketch a Marxian answer to the question posed by Wright’s article in terms I have found useful in attempting to clarify the theory to myself. Whether the theory of surplus value is right or wrong is a separate question, with a long and contentious history, and I don’t hope to prove or disprove the theory here. In any case, the only final test of the theory is how well it describes reality.
The above examples point to the significance of Marx’s discussion in Capital of the dual character of use value and exchange value. Here, it is worth quoting at length from an illuminating article by Nick Beams, from 2003, “A letter and reply on the “Sraffa-based” critique of Marx”:
Analysing the social process of exchange and what it tells us, Marx explains that, first of all, in each of the two different things (corn and iron) something must exist which is common to them both, but which is neither one of them.
As he goes on to demonstrate, this common something cannot be a natural property, because such properties attract our attention only insofar as they comprise the commodity’s use value. “But the exchange of commodities is evidently an act characterised by a total abstraction from use-value.” [Emphasis added]
In my view, this passage is extremely important. Marx does not say, as do the bourgeois economists in constructing their models, that we abstract from this or that, or we make this or that assumption. Instead, he draws out that it is the “act” of exchange that abstracts from use-value and, consequently, from the concrete labour which produced that particular use value. This means that the concept “abstract labour” is not an assumption that we somehow impose. Rather, it is a correct reflection, in thought, of an objective social process.
For a long time, I thought that Marx’s very first argument in Capital, in which he makes an equation of two commodities in order to find a common “substance,” was a poor argument and even idealist, but as Beams points out, Marx is not making the equation here. The equation is made on the market, and this implies that the equality of the commodities cannot be based on their use values, but only on the process of exchange, which in turn leads Marx to posit that their common social “substance” (i.e. what causes them to be regarded as equals on the market) can only be the commodity common to their production, namely abstract labor. As Beams clarifies here, Marx does not introduce the abstractions of exchange value and abstract labor as a matter of mere definition. Rather, an abstraction has been made in the process of production, and in that sense really does exist — not in a Platonic realm, but in the social relations underlying commodity exchange.
Gil Schaeffer’s Critique of Wright’s Value Theory
Gil Schaeffer, in his response letter to Wright’s piece, correctly points out that Wright’s theory is not Marx’s, and rightly points to the critical question being that of control over the means of production: “The question isn’t whether humans have creative powers that machines don’t, but who controls and directs that creativity and for what purposes.”
Yet Schaeffer goes on to level the very same criticism at Marx that prompted Wright’s “causal powers” theory in the first place: namely, that value theory is tautological because Marx defines labor as the only source of value and claims without justification that machines pass all of their value to their products while only labor adds something. Thus, Schaeffer claims that Marx “keeps two sets of accounting books, one for machines and one for labor.” As we have already seen, the equation of labor with value is not, for Marx, a mere definition, but comes out of a model of market relations and unequal power between the wage laborer and the capitalist.
The question is not, as Schaeffer puts it, whether the machine can “add to the productivity of existing labor over and above its cost for it to be worthwhile to the capitalist to invest in it.” Of course, this must be the case, or economic output would never grow. Does Schaeffer imagine that Marx wasn’t aware of this fact? Schaeffer merely repeats Wright’s confusion of a surplus product with surplus value. He imagines profit from the perspective of an individual capitalist, as the difference between the cost of the outputs and the inputs, rather than considering how profit is determined on the whole market. He treats productivity and cost as fixed categories of one section of capital, but these categories have no meaning except with respect to the rest of the market.
Whereas Marx starts from the circuit of social production and exchange, bourgeois economics starts from the perspective of one section of capital, because this is what is immediately useful to the individual capitalist. However sophisticated the mathematics arising on this basis, this perspective is neither interested in nor capable of modeling the development of capitalism as a historical system. Hence, Schaeffer misses that while machines may add to the productivity of labor, the capitalist that owns them does not produce for use value but only for exchange value. Machines can be a source of profit only insofar as they produce exchange values, and to that extent, they can be treated as constant capital because they themselves are commodities that are subject to competitive markets.
Once again, both the original article and the response letter misunderstand Marx and tail bourgeois criticisms on two key points:
- That the equation of labor with value is a tautology.
- The confusion of an increase in wealth or productivity with profit. (And similarly, in the confusion of surplus product with surplus value, use value with exchange value, and concrete labor with abstract labor.)
Once again, the validity of the theory is a bigger question than we can answer here, but let’s not presume, as in Schaeffer’s response letter, to “dispose of” Marx’s theory with the wave of a hand on the basis of basic misunderstandings.
Schaeffer asks: “But do workers really need to know complex economic theories in order to be convinced that the capitalists hold state power, write and enforce the laws of property to their own benefit, and call on the police to enforce their rule in case it is challenged?” Here, we can agree with Schaeffer that workers do not need to understand the intricacies of Marxian economics in order to carry out a revolution, because they are driven in the first place by their needs and the development of the productive forces, not by theories. But the theory is a powerful weapon in the hands of the working class in cutting through the pervasive apologetics for capitalism.
The core of Marx’s theory is the class struggle: that the capitalist class, by its control of the means of production, appropriates the profits; that the working class, meanwhile, lives on wages that are driven toward subsistence levels by the market; hence, that “equal exchange” masks coercion, and no matter how much technology advances, no matter how much wealth humanity piles up, the working class does not share in the profits. Rather, it is more and more shoved aside and subordinated to a growing mass of machinery. The more the capitalist system advances, the more glaring the contradiction becomes between the huge surplus that can be produced on the one hand and the inability to satisfy human need on the other. This much will be apparent to the working class, and will become more apparent as the AI revolution picks up speed.
All of this can be recognized without the theoretical work of Marx, which is aimed at uncovering the inner workings of this process and its historical development. Yet the theory is of further importance in exploding the claims of bourgeois apologists that capitalism’s problems can be reformed away, or that capitalism is the realization of eternal laws of nature (i.e. natural selection) in the economic realm, hence never to be transcended. As Paul Mattick writes:
Marx’s abstract theory of capital development, though incapable of predicting the definite end of capitalism, is significant as an instrument to be set against the persistent illusion that capitalism could actually reach that state of tranquility held out by its apologists as the only hope of the future. It helps explain why all the concrete contradictions encountered in reality cannot be considered accidental or remediable shortcomings. These difficulties, singly and as a developmental pattern, are due to a trend in capital accumulation itself. When capitalism’s inner connections are grasped, Marx wrote, “all theoretical belief in the permanent necessity of existing conditions breaks down before their practical collapse.”3
There is thus enormous incentive for bourgeois economists to reject Marx’s theory, because much of the purpose of their field is to defend the rationality of the capitalist system, and to treat it as an ahistorical system or algorithm that can endure indefinitely. A theory that posits that capitalism cannot overcome its problems but is driven by its own development to hurl itself apart cannot be accepted by them.
As Schaeffer notes, theorizing is no substitute for class struggle, but ideology is a component of the class struggle, albeit a subordinate one, so we should give these theories the respect and careful consideration they deserve.
Those who see the struggle for socialism as a purely political matter will naturally see economic theory as unimportant. But capitalism cannot be overthrown by a party, or the struggle for a democratic republic, isolated from the wider struggle of the working class to replace capitalism with a new economic order. The political struggle is only one manifestation of the wider class struggle.
AI, the “Absolute Inner Limit” of Capitalism
Set against the specter of AI and the spreading automation of the economy, Marx’s theory implies that the very process of accumulation, the motive force of growth under capitalism, will lead to the breakdown of the capitalist system. The falling rate of profit is the quantitative expression of the growing productive forces coming up against the barriers of the social forms of production. In the limit of a fully automated economy, with no input of human labor, the profit rate would fall to zero, which clearly signals the breakdown of the capitalist market.
Ernest Mandel, in Late Capitalism, concludes for this reason that full automation would be “the absolute inner limit of the capitalist mode of production.” Mandel writes, however, that “capital must by its very nature put up growing resistance to automation beyond a certain point”, and claims that “…under capitalism full automation, the development of robotism on a wide scale, is impossible, because it would imply the disappearance of commodity production, of market economy, of money, of capital and of profits.” But the resistance of a declining mode of production to a rising one has its limits. Given sufficient time, automation will inexorably spread throughout the whole economy, because of the competitive advantages it provides to each section of capital. If capitalism represents a barrier to the full development of automation, this only means that the “integument” of capitalism must eventually be “burst asunder.”
Mandel points to this quote of Marx:
A development of productive forces which would diminish the absolute number of labourers, i.e., which would enable the entire nation to accomplish its total production in a shorter time span, would cause a revolution, because it would put the bulk of the population out of the running. This is another manifestation of the specific barrier of capitalist production, showing also that capitalist production is by no means an absolute form for the development of the productive forces and for the creation of wealth, but rather that at a certain point it comes into collision with this development.” (Capital, Vol. III)
The socialist movement has not yet absorbed how revolutionary AI will be. Where Mandel and other past socialists saw the unlimited spread of automation — and the fall of the rate of profit to zero — as only a theoretical possibility, we today confront it as an imminent reality.
If humanity is to transcend the failing capitalist system, the working class will have to wrest automated industry out of the hands of capital, and set it to work as the collective property of humanity, to be democratically owned and administered by the entire population. Whether we end up with socialism or techno-slavery will be up to the class struggle. However, Marx had a firm conviction, based on his analysis of capitalism’s development, that the result would be socialism. Engaging with his economic theories is thus more relevant than ever.
- Footnote: See, for example, his Social Architecture Model. Link: https://arxiv.org/abs/cond-mat/0401053
- This example was motivated by the discussion in Nick Beams’ 2003 article, “Machinery and the origins of surplus value”.
- Paul Mattick in “Value Theory and Capital Accumulation,” citing Marx’s letter to Kugelmann