Overview of NPR’s It’s Been a Minute: “Is your data getting colonized?”
This episode examines the hidden human labor behind artificial intelligence, using Waymo’s remote support workers in the Philippines as a starting point to explore a larger argument: modern AI systems are built on global labor extraction, data labeling, and colonial power dynamics. Brittany Luse speaks with journalist Regine Cabato and scholar Ulises A. Mejias about how AI depends on underpaid workers, how data is sourced and used without meaningful consent, and why some see this as a form of “data colonialism.”
Key Points
The Waymo example revealed the human side of “autonomous” AI
- The episode opens with the surprise that Waymo self-driving cars can route difficult situations to human workers.
- During a Senate hearing, Waymo confirmed that some of those remote workers are located in the Philippines.
- This sparked online reactions and highlighted how “self-driving” tech still depends on human oversight.
AI is far less “artificial” than it appears
- Ulises Mejias explains that AI systems rely heavily on human labor:
- labeling and tagging training data
- supervising systems in real time
- moderating harmful content
- The public narrative that AI works “by itself” hides the fact that people are doing much of the tedious, repetitive, and often invisible work.
Data annotation in the Philippines is a major labor pipeline
- Regine Cabato describes a large crowd-work industry in the Philippines, with workers tagging images and data so AI can learn to identify objects and scenes.
- These jobs are often:
- gig-based or contract-based
- low-paid
- lacking benefits and protections
- opaque in terms of working conditions and pay
- The transcript notes that firms like Scale AI have used this labor for major clients, including U.S. government and tech companies.
The labor model reflects colonial and post-colonial patterns
- Both guests connect AI labor outsourcing to the Philippines’ colonial history with the United States.
- The Philippines is attractive to Western tech companies because of:
- widespread English fluency
- a digitally literate, highly online population
- a long history of exporting labor for U.S.-based firms
- Mejias argues this is part of a broader system he calls data colonialism: extracting human labor, data, and value from the Global South for profit and control.
AI also depends on extracting creative work without consent
- The episode expands beyond labor to include intellectual property concerns.
- Regine Cabato notes that Filipino authors and academics have had their work used to train AI models without permission.
- The discussion compares this to plagiarism:
- not just direct copying
- but also paraphrasing or repackaging someone else’s ideas without attribution
- The central concern is lack of consent and fair compensation.
Content moderation and training can be psychologically harmful
- Ulises Mejias describes workers in Africa and Asia being recruited to review graphic or violent content for AI moderation tasks.
- These workers may be:
- hired under misleading pretenses
- exposed to traumatic material for long hours
- paid very little for emotionally damaging work
- The episode underscores that AI’s hidden labor can be both exploitative and harmful.
Main Takeaways
- AI is not a magic, self-sufficient technology; it depends on large amounts of human labor.
- Much of that labor is outsourced to the Global South, where companies can pay less and avoid scrutiny.
- The Philippines is a key site in this system because of its colonial history, English-language workforce, and digital labor infrastructure.
- “Data colonialism” describes how modern tech companies extract data, labor, and value in ways that resemble older colonial practices.
- Responsible AI would require:
- transparency about who does the work
- fair pay and labor protections
- consent from creators and workers
- stronger regulation
- sustainability and reduced environmental cost
Notable Insight
“It’s very much a Wizard of Oz type of situation: they sell this magic figure, but if you pull back the curtain, it’s just a human behind it.”
That line captures the episode’s central argument: AI’s impressive outputs often depend on hidden human workers whose labor is undervalued, obscured, and extracted.
Why It Matters
This episode reframes AI from a story about futuristic innovation into one about labor, power, and inequality. It asks listeners to consider not just what AI can do, but:
- who makes it work,
- who pays the cost,
- and who benefits from the illusion of automation.