The People Behind Your AI
Subtitle: A UN report pulls back the curtain on the millions of workers who make AI possible, and the conditions they work in.
Key Takeaways
- Every AI system you use (ChatGPT, Siri, Netflix recommendations) was trained and refined by real human workers.
- Content moderators, many of them women in developing countries, review hundreds of violent or disturbing images and videos every day to keep AI safe.
- Gig and delivery workers are increasingly “managed” by AI algorithms, with no human boss to appeal to.
- A joint session by the UN’s International Labour Organization (ILO) and International Telecommunication Union (ITU) on March 3, 2026, highlighted these growing concerns.
- As phones and devices get smarter, the hidden workforce behind them keeps growing, largely unseen.
When you ask ChatGPT a question, tell Siri to set a timer, or scroll through recommendations on YouTube, it feels like magic. A machine just… knows things. It understands you. It responds in seconds.
But here is something most people never think about: behind every one of those interactions, there were real people doing real work. Not just the engineers who built the software, but thousands of workers around the world who trained, tested, and refined it. And according to a recent United Nations report, many of those workers are not being treated well.
How AI Actually Learns
Let’s start with the basics. AI doesn’t learn on its own. It needs examples. Millions of them.
Think of it like teaching a child to recognize a dog. You show them pictures: “This is a dog. This is also a dog. This is a cat, not a dog.” Over time, they figure out the pattern.
AI works the same way, except instead of a parent showing pictures, it is workers (often hired through outsourcing companies) who label data, answer questions, rate responses, and flag problems. They are the ones who teach the machine what a good answer looks like, what is offensive, what is accurate, and what is nonsense.
Every “smart” reply your AI assistant gives you started with a human being deciding whether a similar reply was helpful or not.
The Human Cost of “Safe” AI
One of the most striking findings from the UN session is about content moderation. To keep AI systems from generating violent, hateful, or sexually explicit content, someone has to show the AI what that content looks like, and then label it as harmful.
That means real people sit in front of screens, watching hundreds of disturbing videos and images per day. Graphic violence. Child exploitation. Hate speech. Their job is to categorize it so the AI learns to filter it out.
Many of these workers are women. Many are in developing countries like Kenya, the Philippines, and India. The pay is often shockingly low, sometimes just a few dollars per hour. And the psychological toll is enormous. Studies have documented PTSD-like symptoms, anxiety, and depression among content moderators. Some companies offer counseling. Many do not.
Here is the uncomfortable truth: every time an AI system successfully blocks a disturbing image or refuses to generate harmful content, it is because a human being somewhere already absorbed that harm on your behalf.
Managed by Machine
The UN report also highlighted a different kind of AI labor problem: workers who are not training AI, but being controlled by it.
If you have ever ordered food through a delivery app, the person who brought it to your door was likely managed entirely by an algorithm. The app decides which orders they get, how fast they need to deliver, and how their performance is rated. If the algorithm decides they are too slow or get too many complaints, they can be deactivated (essentially fired) with no human review and no appeal process.
This is not a future scenario. It is happening now, to millions of gig workers worldwide. Ride-share drivers, warehouse workers, freelance content creators. The ILO and ITU joint session on March 3, 2026 specifically called attention to this trend, noting that algorithmic management is expanding faster than the regulations meant to govern it.
Imagine being fired by a formula, with no manager to talk to, no HR department to call, and no explanation beyond “your performance metrics were insufficient.”
The Invisible Workforce Keeps Growing
This issue is not getting smaller. It is accelerating.
Samsung’s latest Galaxy S26 lineup includes on-device AI features that can summarize your day, manage your schedule, and even draft messages in your voice. Google’s Pixel phones are rolling out “agentic” features that can take actions on your behalf, like booking appointments or comparing prices. Apple continues to expand Siri’s capabilities with every update.
Each of these features requires training data. Someone has to label the voice clips, rate the summaries, and verify the recommendations. As our devices get smarter, the hidden workforce behind them grows. Yet most consumers have no idea these workers exist, let alone what they are paid or how they are treated.
What You Can Do
You do not need to stop using AI. And you should not feel guilty every time you talk to Siri. But there are a few practical things worth considering:
Ask questions. When a company launches a new AI feature, look for information about how they source their training data and treat their workers. Some companies (like Anthropic and Google) have published transparency reports. Others have not. The difference matters.
Support better standards. Organizations like the Partnership on AI and the ILO are pushing for fair labor standards in AI supply chains. Following their work, sharing it, and supporting the conversation helps move things forward.
Stay informed. The more people understand how AI actually works (not the magic version, but the real version), the harder it becomes for companies to ignore the human cost. Read the reports. The UN session documents are publicly available at news.un.org.
Talk about it. Mention it to friends. Not as a lecture, but as something genuinely interesting and worth knowing. Most people are surprised to learn that AI has a human workforce at all. Awareness is the first step.
The Bottom Line
AI is not built by machines alone. It is built by people, millions of them, doing difficult, often invisible work so that the rest of us can enjoy smoother, smarter technology. The UN’s ILO and ITU are now formally asking the world to pay attention to these workers and the conditions they face.
That does not mean AI is bad. It means AI is a product, like anything else, and the people who make it deserve fair treatment. As these systems become more central to daily life (and they will), the question is not whether we use AI. It is whether we care about the people behind it.
The answer should be yes.
Sources: UN News, European Sting, UNICEF Innocenti Office of Research