Rest of World reported that AI companies are hiring highly educated workers across the Americas to label data and provide feedback for models that could eventually automate those same professions. In São Paulo a radiologist earned $30 per hour annotating chest X‑rays for a startup building a diagnostic AI; in Mexico City a contract lawyer received $25 per hour tagging legal clauses for a legal‑tech platform. These tasks are part of the reinforcement learning from human feedback (RLHF) pipeline used by firms such as Scale AI, Appen and Lionbridge to improve large language models and medical imaging systems.

The market for AI training data is expanding rapidly. According to Grand View Research the global data labeling market will reach $1.2 billion in 2024, with the high‑skill segment growing at 35 % year‑over‑year. Scale AI alone raised a $1 billion Series F round in early 2024, pushing its valuation to $13.8 billion. The payments described in the article represent a small but visible slice of that spending, where firms pay premium rates to capture nuanced expertise that generic crowd workers cannot provide.

Paying experts to train their own replacement creates a dangerous externality. It accelerates the very displacement it pretends to mitigate, turning professional judgment into a commodified input that enriches platform owners while eroding the long‑term earning power of the workers who supplied it. This dynamic concentrates value in a handful of AI infrastructure firms and threatens to hollow out the knowledge‑based economies of the Americas.

Second‑order effects include a talent drain from core professions as skilled individuals divert hours to labeling work that offers no career progression, reduced incentives for deep skill development, and a growing likelihood of regulatory backlash. Lawmakers in the U.S. and Brazil are already scrutinizing labor practices in AI data labeling, and professional associations may push for collective bargaining or data ownership rights to protect their members’ livelihoods.

Winners are the data labeling platforms—Scale AI, Appen, iMerit—and the tech giants that acquire the resulting models, notably Google, Microsoft and Meta. Losers are the highly educated workers in the United States, Brazil, Mexico and other Americas nations whose expertise is being extracted for low‑wage, short‑term gigs, as well as small law firms, clinics and engineering consultancies that lose competitive advantage when their specialized knowledge is encoded into broadly available AI systems.

truncated