The Rise and Fall of the Knowledge Worker
Knowledge workers, whose jobs involve handling information rather than producing goods, were supposed to be the beneficiaries of neoliberalism and globalization. But generative AI and a hypercompetitive employment market is immiserating them too.
A former coal miner works at a computer station at the Bit Source LLC office in Pikeville, Kentucky, on February 1, 2016. (Sam Owens / Bloomberg / Jim Ratliff / Getty Images)
At a recent gathering of corporate leaders and US government officials hosted by the venture capital firm Andreessen Horowitz, Vice President J. D. Vance presented a strikingly candid analysis of the last fifty years of US economic policy. “The idea,” he said, “was that rich countries would move up the value chain, while poorer countries handled the simpler tasks.”
What he meant by this was that since the 1970s, supporters of globalization assumed that while some workers in places like the United States might lose manufacturing jobs, most would adapt. They would, to use a phrase that became a meme in the 2010s, “learn to code.” By swapping coalfaces for laptops, workers in the United States, where high-value jobs would be concentrated, would occupy a higher position in the global value chain than their counterparts in the Global South. What happened instead, Vance lamented, was that “as they got better at the low end, they began catching up at the high end too.”
Vance’s description of this tendency is, in a sense, more honest than what the world has come to expect from American politicians. Since the Cold War, American leaders have sold globalization in the slick idioms of progress, integration, and modernization — a form of trickle-down economics for nation-states that would both further enrich the wealthy and uplift the “underdeveloped.” And while standards of living have indeed risen since then — most drastically in East Asia — the reality for the rest of the world has been middling growth, accompanied by the disastrous collapse of state institutions and welfare.
Railing against the evils of globalization, Vance posits a world shaped by a zero-sum race for supremacy between nation-states. Missing or conveniently elided from this narrative, however, is a serious analysis of class, despite it being the primary axis determining who benefits from globalization. Lumped together under the name of the nation are exploiter and exploited, those ruthlessly seeking to maximize their profits across industries and geographies, and those bearing the brunt of this insatiable drive for accumulation.
Presenting themselves as champions of the American working class, Vance and politicians like him deflect attention away from their billionaire sponsors toward foreign workers and a vaguely defined liberal urban elite, largely drawing on the cleavage between blue-collar and white-collar workers.
Fordism and Post-Fordism
The economic system for which Vance and other members of the populist right are nostalgic is what is often called the Fordist era of capitalism. During its heyday, the so-called golden age of capitalism, roughly one in every six American workers was employed, directly or indirectly, in the auto industry; today, the figure is just under 3 percent.
Fordism was defined by mass consumption across society and mass production in factories organized according to Taylorist principles of hyperstandardization of work methods, tools, and equipment to maximize efficiency. It represented a particularly successful period of capitalist growth. In the United States, for instance, between 1947 and 1979, average wages for workers in nonsupervisory roles rose by 2 percent per year, while real GDP grew by 7.3 percent. In comparison, from 1979 onward, wages grew by only 0.3 percent per year, while real GDP grew by just 4.9 percent.
The demise of Fordism, which began in the 1970s, was brought on by intensifying international competition. Other advanced capitalist countries, such as West Germany and Japan, began to produce goods similar to those of the United States. Lower wages in those countries, combined with the duplication of productive capacities, ultimately exerted downward pressure on prices — and, eventually, profits.
The effects of this collapse were expressed in changes to both the production of goods and the consumption patterns of Americans. Lean factories, coordinated by increasingly complex globalized supply chains, replaced the mass domestic manufacturing of standardized goods. Advances in automation, computing, and communication technologies facilitated this transition by enabling the management of a more flexible and geographically distributed workforce.
People’s consumption patterns also changed: ordinary Americans gained access to a wide range of increasingly individualized commodities at cheaper costs, from diverse items of clothing that were tailored to emerging subcultures, to infinitely personalizable Funko Pops. This mode of consumption soon grew to become the aspirational norm for the middle classes all over the world.
But the decline of Fordism also led to the erosion of the labor movement in most of the Global North. The proximate cause was the offshoring of factories and the mass layoffs of organized workers. As these workers were displaced into the smaller, more spatially dispersed workspaces that the service sector demanded, their capacity to organize became more constrained.
This period eventually led to crushing defeats for unions, and erstwhile productive hubs — the American Rust Belt, the north of England, northern France — saw rapid deindustrialization as factories moved abroad, aided by standardized shipping containers, computerized inventories, faster communication networks, and a number of other technological innovations.
This created a Cartesian split within the global economy between a Northern mind, where intellectual, creative, and managerial labor was performed, and a Southern body, responsible for the production of physical goods. The North American Free Trade Agreement (NAFTA), a free trade agreement between the US, Canada, and Mexico, signed in 1994, and China’s entry into the World Trade Organization in 2001 exacerbated these trends. Production was mostly relocated to Asia, initially to South Korea and Taiwan, and eventually to mainland China.
There, nations with large peasant populations and innovative forms of hands-on governance offered both ever-ready pools of labor and rigid labor discipline. China, for instance — by far the largest example of such a manufacturing hub — ushered in what has come to be known as the dormitory labor regime, which clustered workers in dense accommodation at their place of work, affording factory management unprecedented control over their employees’ daily routines.
While a small group of underdeveloped East Asian nations was able to benefit from globalization, the vast majority of countries assimilated into these networks — from Egypt to South Africa to Indonesia — saw the erosion of state capacity and welfare under the discipline of financial capital and remained trapped in low-value services and petty commodity production.
The Rise of the Knowledge Economy
At the same time, rapid improvements in computing and communication technologies helped birth a new class of knowledge workers: data modelers, software developers, system designers, financial analysts, and network engineers. This new class served as the middleman for increasingly discongregated flows of capital, resources, information, and commodities. Members of this class enjoyed relative stability by receiving larger shares of corporate profits, either directly through higher wages or through their ownership of equity. This subset of labor became the managers and facilitators of post-Fordist capitalism and saw their standards of living and capacity to consume comfortably rise.
In the minds of advocates of globalization, these new jobs were supposed to offset the losses resulting from deindustrialization. Yet the gains distributed by these jobs were highly unequal, with a small sector of upper-income households reaping most of the benefits: the Gini index for income inequality in the United States, for instance, grew from 0.45 in 1971 to 0.59 in 2023, a level last seen only prior to World War II.
In North America, this upper crust of workers thus reaped most of the benefits of globalization; in Europe, higher taxation somewhat mitigated this divergence, redistributing part of the gains made by the new middle classes to a broader class of workers through what remained of the welfare state. But in reality, both models were rather disconnected from where a large proportion of profits were being generated — in the factories of China and Mexico, and the mills of Bangladesh and Vietnam.
Emblematic of this new economy is the Swedish fashion retailer H&M. In 2024, the company posted an operating profit of $1.8 billion. It paid an average tax rate of 24.9 percent, with virtually none of it in Bangladesh, where around 20 percent of its garments are produced. A clothing designer at H&M can earn up to $100,000 a year, while the monthly minimum wage for a textile worker in Bangladesh was only recently raised to $113: a measly $1,356 per year.
Generative AI and the Inward Turn of Capital
In recent years, the small subset of workers who have benefited from the globalized economy have started to feel the pinch. The rise of generative AI and the widespread anxiety about its effects can be read in this light. Since ChatGPT’s launch in November 2022, it has become increasingly clear that myriad forms of labor — graphic design, copywriting, programming — are rapidly being subjected to the same logic of discipline that was once centered in the factory.
While much unjustified hype has accompanied generative AI, and the technology is indeed far from perfect, its capacity to write computer code or generate product design and marketing imagery is rapidly improving. It is no longer entirely unreasonable to conclude that something akin to a process of industrial proletarianization might gradually reach forms of informational and creative labor that had thus far been immune from these shifts.
Even if we don’t accept fantastical notions of artificial general intelligence (an AI that could surpass human intelligence), or grand declarations of a fourth industrial revolution, in its current form, generative AI models are capable of helping capitalists impose wage discipline upon a broad range of knowledge workers. Their capacity to efficiently search and process large volumes of text poses a particular threat to occupations built around the discovery, curation, and organization of knowledge.
These models have also been deployed to automate certain aspects of software development and computer programming, de-skilling programmers and reducing the leverage they once held. A generative language model, for instance, can now produce most of the code scaffolding needed for a reasonable prototype of a website or mobile app within an hour or two — work that would typically take the average software developer a few days. In domains such as marketing, content creation, and advertising, generative AI models are capable of supplanting a large portion of an employee’s tasks. Whether they do so well is beside the point: little prevents market forces from turning “AI slop” into the new standard.
Decline of the Aristocracy
The success of the philosophers Michael Hardt and Antonio Negri’s Empire at the turn of the millennium sparked a renewed interest in a strand of contemporary labor analysis that had been particularly popular among Italian Marxists since the 1970s. These so-called post-workerist thinkers — such as Maurizio Lazzarato, Paolo Virno, and Negri himself — argued that the informational, cultural, and communicative forms of networked labor were more resilient to measurement and less susceptible to absorption into circuits of discipline and commodification. In immaterial and cognitive labor, they saw the seeds of autonomy, cooperation, and the potential for post-capitalist forms of production — in other words, a form of liberation from exploitative labor itself.
In hindsight, these ideas were ultimately rather out of step with the reality of how these “immaterial” patterns of labor ended up evolving. As with other recent developments in different types of knowledge work — such as Agile software development or metricized content creation — generative AI serves to expand the logic of the factory to precisely these seemingly autonomous labor patterns, routinizing them and making them more amenable to discipline. A graphic designer, for instance, can now be asked to deliver a 3D model in an hour instead of a day by an employer who can tell them to use Midjourney or any such AI assistance tool.
Today capital’s web is shrinking. The mesh connecting microchip producers at Foxconn’s factories in Shenzhen to Genius Bar employees in Berlin to the tech workers at Apple’s Cupertino offices is growing more uniform. While the position of the low-end and high-end worker vis-à-vis capital is vastly different, they increasingly share a downward trajectory.
In what is a telling sign for the tech sector, computer programmer employment rates in the United States have plummeted to their lowest level since the 1980s. Such pressure has visibly eroded workers’ capacity to negotiate, and not just over wages. In 2018, Google employees were able to halt the company’s collaboration with the US military under Project Maven. Last year, on the other hand, over fifty workers were met with summary dismissal after protesting Google’s complicity in the genocide in Gaza. The aristocracy of the knowledge economy, once able to negotiate its terms, is slowly being dethroned.
Now more than ever, it is essential for us to fight against the atomization that keeps workers separated across global supply chains. As Northern capitalism’s inward turn accelerates, it becomes more and more crucial to look outward — to cultivate alliances and solidarities with data center engineers, textile workers, platform workers, cobalt miners, and all those relegated to the low-end, to the shadows of global capitalism. Capital is a far more formidable adversary today than it was half a century ago, and if we are to build a successful labor movement, it is crucial that we willfully, deliberately build solidarity and organize across every node in its web.