📨 AI for Social Impact Deep Dive: Data Workers

✍🏼 A Note From the Editor

In this deep dive, we talk about the human intelligence powering AI. Behind every chatbot is not only the training data it ingested to make it smart, but also a global workforce of humans teaching it too. We are going to pull back the curtain on data workers to understand the human training component of AI intelligence.

🧩 What Is a Data Worker?

Data workers are the people who label, annotate, and clean data. In other words, they are the human judgment that AI systems are built on. For example, their job could entail tagging cars and pedestrians to teach autonomous vehicles not to hit them, or circling abnormalities in CT scans and X-rays to teach AI to recognize diseases. They also rate and correct chatbot answers, a process called Reinforcement Learning. What’s more, the global data annotation market is forecast to reach $10.2 billion by 2034.

🇺🇸 The Credentialed Data Workers

In the US, intermediary companies work with Big Tech companies to recruit doctors, lawyers, scientists, and PhDs (often individuals who have limited job prospects due to AI in the first place) to do the work that makes models smarter: stress-testing reasoning, scoring outputs, and labeling tricky edge cases. But these are gig contracts, not full-time jobs with legal protections. Workers can be offboarded abruptly when projects end or quality dips. In September 2025, US workers at GlobalLogic who were improving Google's Gemini models reported being fired after raising concerns about working conditions and low pay.

🌍 Global Data Workers

Outside the US, data workers concentrated in places like Kenya, Venezuela, and the Philippines are subject to traumatizing work like image labeling, toxic-text tagging, and content moderation. Workers in the Philippines employed by Scale AI reported being paid below the legal minimum wage and payments are often delayed or withheld for completed tasks. Kenyan data workers similarly reported a lack of formal contracts with clear terms, jobs that vanish unpredictably when projects are taken offline, and sudden dismissals without warning or recourse.

💰 A Few of the Companies

While there are at least 30 intermediary companies for data work, here are some main players:

Scale AI — Meta paid $14.3 billion for a 49% stake in Scale AI in June 2025, with founder and then-CEO Alexandr Wang stepping down to join Meta. Concerns over Scale's neutrality reportedly caused Google to cut ties.

Mercor — Mercor hit $1 billion in annual recurring revenue in April 2026. It manages over 30,000 contractors who are collectively paid more than $2 million per day.

Surge AI — Surge AI generated over $1.2 billion in revenue in 2024. Then in July 2025 it entered talks to raise about $1 billion in its first-ever funding round at a valuation of at least $25 billion, which would make it one of the most highly valued startups in the US.

Handshake AI - Handshake AI was launched in June 2025. Handshake's platform already had more than 18 million users, three million of whom hold graduate-level degrees and more than 500,000 of whom have PhDs, a ready-made expert pool for frontier labs. Handshake AI’s gross annualized revenue from AI training has surged to nearly $1B, up from $550M in January 2026.

📰 The Good News?

Data workers are organizing!

• Kenya: Data workers have formed the Data Labelers Association, created to improve working conditions and raise awareness about the challenges they face — building on the African Content Moderators Union formed in Nairobi in 2023.

• Globally: The Global Trade Union Alliance of Content Moderators launched in 2025 with workers from nine countries, backed by UNI Global Union, pushing for living wages and safe working conditions across Big Tech's value chain.

• Europe: In January 2026, workers at Covalen's Dublin offices — who provide services for Meta, including AI training — went on strike to demand union recognition, better wages, and improved leave conditions.

And through the Data Workers' Inquiry (from Timnit Gebru's DAIR Institute), data workers can document their own conditions through reports, documentaries, and podcasts. Knowledge is power. ✊

👋🏼 About AI for Social Impact

I’m Joanna, and I’m on a mission to help folks in the social impact sector understand, experiment with, and responsibly adopt AI. We don’t have time to waste, but we also can’t get left behind.

Let’s move the sector forward together. 💫

♥️ Spread the Love

Spread the love and forward this newsletter to anyone who might benefit from a dose of AI inspo!