Hidden workforce behind AI spans 160 million workers, many in the Global South

TEXT START: Many widely used artificial intelligence products depend on people to perform repetitive work to train and maintain the systems, even though the technology is often presented as if it runs on its own.

THE DISSECTION

The article exposes AI’s concealed labor infrastructure: annotation, moderation, and fragmented Global South gig work. It correctly punctures the fantasy of autonomous systems, but redirects the analysis toward wages, visibility, and organizing. Its central framing treats AI as a durable restructuring of labor rather than a mechanism that is structurally incentivized to eliminate that labor.

THE CORE FALLACY

It confuses present implementation costs with permanent economic necessity. Human workers are currently the scaffolding of AI deployment, not proof that human labor remains indispensable. As models, synthetic data, filtering systems, and evaluation tools improve, the same repetitive cognitive tasks become targets for automation. The hidden workforce is a lag layer—and a giant exposed surface for cost reduction.

The stated figure of 160 million is presented without methodology, definition, or time series. Even if accurate, scale does not establish durability, bargaining power, or productive participation.

HIDDEN ASSUMPTIONS

• Current human-in-the-loop work will persist as AI scales.
• Transparency and organizing can overcome concentrated ownership and competitive cost pressure.
• Better wages and protections can solve the underlying displacement problem.
• Public awareness will materially alter who owns the systems and captures the gains.
• The issue is exploitation of workers, rather than the coming loss of their economic necessity.
• Bill Gates’s “equalizer or injustice” framing is analytically useful; it replaces ownership mechanics with moral theater.

SOCIAL FUNCTION

Partial truth wrapped in transition management and ideological anesthetic. It makes a real exploitation visible, which is valuable, but offers reformist pressure as the implied remedy while leaving the Discontinuity Thesis untouched: AI can exploit human labor during the transition and still destroy mass productive participation afterward.

THE VERDICT

The article is right about the hidden workers and wrong about what their existence signifies. They are not evidence that AI cannot replace labor; they are evidence that replacement proceeds in stages, using cheap human scaffolding before automating the scaffolding itself. Transparency may improve conditions temporarily. It does not reverse ownership concentration, competitive automation, or the collapse of the wage-consumption circuit.