Recent hysterical coverage of the AI boom suggests that soon millions of workers will be replaced by chatbots that will do their jobs. This is not the first time these kinds of exaggerated claims have been made. Those making the statements are usually directly invested in AI and its infrastructure. Insane levels of investment from chips to data centres are pouring into the industry even though, to date, not a single profitable product has been produced by any of the big tech involved. Those involved need hype and hysteria to keep the venture capital rolling in.
“Amazon, Microsoft, Google, and Meta are expected to spend more than $700 billion on data centres and AI infrastructure in 2026” according to one report. Big Tech is pushing the AI “revolution” even as numerous reports show marginal or zero productivity gains for companies and with ever increasing incidents of disastrous consequences when the tech is over relied upon. For Trump and the Tech bros, the AI hysteria is too big to fail with the US economy facing recession if it weren’t for the AI boom.
But will this technology replace workers on the scales envisaged or indeed replace and remove the power of workers in the future? Karl Marx can help us understand what is happening!
The tendency of the rate of profit to fall is a Marxist theory stating that as capitalism develops, the rising organic composition of capital – more machines/raw materials (constant capital), relative to labour (variable capital) – causes the profit rate to decrease. Although technology increases productivity, it reduces the base of surplus value generation.
A more layman‑friendly version of the above could be: As capitalism grows, companies use more machines and equipment compared to human workers. Machines don’t create new value—people do. So even though technology makes production faster and more efficient, it also reduces the amount of human labour in each product. And since profit comes from the value created by workers, the overall rate of profit tends to go down over time.
These days, every company won’t stop talking about AI and all the “amazing efficiencies” it is supposed to bring.
But if we go by the idea of the falling rate of profit – where replacing workers with machines tends to shrink the source of profit – then a question pops up: Are capitalists being foolish by pushing more and more AI into production? After all, profit is their holy grail, right?
The Marxist theory explains how systemic crises of boom and bust occur under capitalism, but Marx also explains how individual capitalists are compelled to invest, innovate, try to create and seize new markets. Ever searching for greater profits, the individual actions of each company make sense for them, but drive the global economic system toward crisis after crisis. So mass layoffs can be threatened in some sectors even as the AI used to replace this work compels millions of low paid workers into service across the globe.
Rather than accepting the hype of the AI industry, socialists need to look closer at what lies behind the curtain of the promised magic AI future; at the workers behind AI. Far from the fantasies of tech billionaires, the working class and its potential power is not being replaced by robots.
The “workers behind AI” include both the foundational pioneers of AI research and a large, often invisible global workforce that labels data, moderates content and refines AI models through microtasks. This latter group, sometimes called “ghost workers,” is crucial for training AI systems but frequently faces low wages, poor working conditions, and a lack of transparency and security regarding their labour, despite their essential work.
Apple boasts of its 1 million ‘flexible contributors’ in more than 170 countries, who focus on more specialised forms of data work that are critical to the “AI supply chain” for “clients including Amazon, Google and Boeing.” As Jathan Sadowski notes, “Through their microwork, this reserve army of labour is crucial to the development of AI, whether because they produce and verify datasets used to train AI systems or because they simply impersonate AI services.” 1The Mechanic and the Luddite; Dr. Jathan Sadowski
Millions of Invisible Workers Behind AI
AI is hardly developed in the fancy offices of Silicon Valley. As reported in a research-based article in Neoma Magazine, far from the sophisticated, sentient machines portrayed in media and pop culture, so-called AI systems are fuelled by millions of underpaid workers around the world, performing repetitive tasks under precarious labour conditions.
Millions of unseen workers keep the whole AI system running with their effort, their time, and too often, their suffering. AI isn’t making work disappear—it’s breaking it into smaller pieces, hiding it behind screens, and making it far more unstable. Behind every chatbot, every image generator, and every shiny “AI breakthrough,” there are countless people doing the hard, repetitive, low‑paid tasks that make these systems possible. They’re invisible in the final product, but without their labour, AI simply wouldn’t work.
The World Bank estimates there are between 154 million and 435 million data workers globally, many of whom are involved in the “hidden” labour that powers AI.
While replacing workers in one place, AI relies on them just as much in a different place;
“much of the cognitive labour essential to their operations likely comes from an office building full of (low-waged) workers in popular outsourcing destinations like the Philippines or India or Kenya” and also increasingly from millions of precarious low paid workers in the Global North.2The Mechanic and the Luddite; Dr. Jathan Sadowski
AI will change jobs more than it will erase them. There is no evidence of immediate mass unemployment. The impact of AI on jobs and working conditions will certainly be uneven across the world and different sectors.
AI will not change the fundamentals of labour–capital antagonism, in other words, the builtin conflict between workers (labour) and owners of production (capitalists) under capitalism. That is not to say it won’t intensify the conflict at the heart of capitalism: bosses want lower costs and higher output; workers want dignity, better wages, and security.
It is essential that workers’ organisations and trade unions build strategies to tackle the AI driven attack on working conditions. Stronger unions in AI-driven workplaces is certainly a good start. We need to reject the idea that mass layoffs and replacements of, for example, health care workers with chatbots are inevitable or that this represents some kind of “progress”.
Key Roles and Types of Workers
AI Pioneers: Individuals like John McCarthy, Alan Turing, Marvin Minsky, Allen Newell, and Herbert A. Simon are considered the founding fathers of AI for their foundational work in computer science. Modern AI development is driven by scientists often called the “Godfathers of AI,” alongside researchers specializing in areas such as mathematics and computer science.
Data Labellers and Annotators: These workers meticulously label, categorize, and annotate vast datasets that are essential for training AI models, from recognizing objects to predicting text.
Content Moderators: Humans who review and filter content are vital for large language models, helping to mitigate toxicity, bias, and the spread of disturbing material.
Quality Raters: These workers evaluate and provide feedback on AI outputs, acting as a quality control layer for the technology.
AI Engineers and Data Scientists: These roles are more visible, focusing on designing, building, and managing AI software applications and analysing data to identify trends
The working conditions of AI workers
The conditions AI workers are subjected to in order to make AI systems run smoothly are grim. Many data labellers and content moderators, especially in the Global South, are paid extremely low wages for their work, which can be outsourced through digital platforms.
The workers are often exposed to traumatic, violent, or disturbing content during content moderation and data labelling tasks.
For example, Open AI is reported to hire Kenyan workers on less than $2 an hour to make Chat GPT content “less toxic”.
There is a lack of transparency and security in the work, with many workers unaware of which company they are working for or what specific AI system their work is contributing to. They may also be required to sign non-disclosure agreements that prevent them from speaking about their work.
The jobs are also extremely precarious. Tasks are often broken into microtasks, and workers can be easily replaced or have their work rejected without recourse, leading to a sense of being “expendable”.
AI work involves a significant amount of unpaid labour, with workers reporting that crucial training and administrative tasks are often unpaid, driving down the effective hourly wage below minimum wage standards.
The International Labour Organisation (ILO) reports: “Artificial intelligence (AI) is often presented as a revolutionary force poised to automate vast swathes of the economy, displacing workers, and ushering in a “post-work” era. However, behind the sleek interfaces and impressive capabilities of many AI systems lies a hidden workforce of humans.”
According to a recent Guardian report, Women in India’s rural communities describe trauma of moderating violent and pornographic content for global tech companies: As one worker said, “In the end, you feel blank’: India’s female workers watching hours of abusive content to train AI”
Working conditions in India’s IT companies are increasingly characterized by widespread burnout due to long hours, pressure for constant availability, and a poor work-life balance. Surveys indicate many professionals work significantly more than the legal 48-hour week and feel obligated to respond to work communications outside of office hours.
Millions of workers are paid peanuts and working in hard conditions to train AI models. Exploitation in the Global South is the backbone of promises of massive profits by AI in the U.S. and Europe. But the exploitation is not limited to the Global South.
As quoted in Global Education Network, Sociologist Antonio Casilli hits the nail on the head: artificial intelligence is, in reality, intelligence of “sweat and tears.” Algorithms don’t learn by themselves; they simply reproduce patterns based on the previous work of millions of people.
Workplaces Without Workers?
AI can automate many tasks, but it cannot create fully worker‑free workplaces, because real-world production, services, and society still depend on human labour in multiple ways. Even AI systems rely on massive amounts of human labour. Behind every “automated” system are millions of workers. Drones, automated cars, chatbots etc, all have armies of real labour and workers behind them. Physical production still needs people.
Capital wants to reduce labour — but cannot get rid of it. Employers will try to use AI to cut costs, deskill work, replace stable jobs, expand surveillance etc but a totally worker‑free workplace would remove the source of value creation and oversight. Even highly automated firms need human labour somewhere in the chain – either on-site or outsourced globally.
AI intensifies the struggle over control but does not remove labour from the system. For the ruling class, AI holds out a promise of finally eliminating its mortal enemy; the working class. But this fantasy is hundreds of years old, as old as capitalism itself. It is an unrealisable fantasy as ultimately all of its profits come from those workers. AI does what every new technology has done; it intensifies the human labour needed and changes the kind of labour required, but workers remain central to the production of surplus value for capitalists. As Sadowski says;
“In other words, the role of human labor has changed because of technology while also becoming more central to the production of value. Behind all that automation is a bunch of people designing, building, using, overseeing, maintaining, repairing, and taking over those tools as needed. The technologies are vessels and conduits for human labor”.3The Mechanic and the Luddite; Dr. Jathan Sadowski
AI will not replace the central role of workers or their potential power. But capitalists will try to use it as they have every tech advance in the past; to discipline, surveil, and make their lives more precarious. New forms of organisation and new fights against this will break out. In the meantime we need to reject the idea that Silicon Valley’s dystopian vision of the future is unchallengeable.
Like the Luddites in early 19th century England we should resist the attempt by elites to force a technology on us that seeks only to worsen conditions for most of humanity and impoverish workers while creating grotesque levels of inequality and environmental destruction. The Luddites were not resisting progress – they fought an attempt by the ruling class to use new technology to deskill and impoverish workers for the benefit of the elites. Today we face a similar misuse of technology, but unlike the Luddites, today’s global working class has greater capacity than ever to fight and resist these attacks
We can start by opposing every data centre they want to build that will worsen the climate crisis and misuse precious resources like fresh water. We should fight every attempt to replace a teacher or care worker with a chatbot. And we should remain highly sceptical of the claims for a technology that to date has consumed vast financial and human resources but delivered very little other than increased surveillance, intensified work, and an internet awash with slop, while deepening inequalities and amplifying racism and fascist narratives.4Resisting AI; An anti-fascist approach to Artificial Intelligence; Dan McQuillan The future is unwritten, but it isn’t going to be some AI sci-fi billionaire’s dream.
