Why the AI data annotation side hustle is booming

Amelia is 24, holds a degree from Oxford, and two years ago she moved to London to start a master's. She missed the deadline for tuition-fee financial support, so she took shifts in a restaurant and then a factory while applying for everything within reach. Nothing came back. The one thing still hiring was the AI data annotation side hustle, she told The Times: cleaning up data so models could learn from it. She has her doubts about AI, morally speaking, but the bills did not wait.

The job she landed has a plain name: data cleaning. Workers categorize, label, and test images, audio, video, and text before any of it can be fed to a model. The models' own outputs then go back through the same human check so they can be improved. It is the invisible labor under every chatbot answer, and the AI data annotation side hustle is hiring at a pace that looks like a boom market.

Postings are exploding

The numbers come from LinkedIn, reported by The Times this week. In Britain, postings for data annotation work rose 250 percent year over year, making the country the fastest-growing market on the planet for the AI data annotation side hustle. India is close behind with postings up 144 percent, followed by the European Union at 96 percent and the United States at 76 percent. For a kind of gig work that barely had a name a few years ago, the growth is remarkable.

The map underneath is lopsided. The United States holds 37 percent of the world's data annotation postings, India holds 32 percent, the European Union holds 7 percent, and Britain, despite growing fastest, holds just 5 percent. Growth is fastest where the base is smallest, which is a familiar pattern in labor markets that are still taking shape. In the United States alone, LinkedIn estimates cited by The Wall Street Journal put roughly 282,000 new data annotator positions on the books between 2024 and 2026, part of more than three-quarters of a million AI-related American jobs created since 2023.

The pay is real, and so is the grind

Entry-level annotators typically earn between $15 and $20 an hour, according to industry salary roundups, while specialists with coding, science, or language expertise can reach $40 to $65 an hour on premium platforms. The spread explains the rush: AI-literacy demand in American job postings rose 70 percent year over year, and the typical AI job posting advertises about one hundred seventy-seven thousand dollars a year against roughly eighty thousand for a non-AI role. Labeling images for a chatbot does not pay six figures, but the pipeline it belongs to is the best-paid corner of the labor market, and the AI data annotation side hustle sits squarely in it.

The catch is what it has always been with gig work. Hours are inconsistent, contracts are often project-based, and the legitimate postings sit in a market thick with scams and lowball offers. Career guides warn that finding real annotation work means vetting platforms carefully, since the good gigs are buried under noise. The oldest large-scale version of this kind of microtask work was Mechanical Turk, which Amazon shut down after 21 years; what has changed is that the tasks now teach models instead of just moving data around. Amelia's ambivalence captures the rest of it: she works in the engine room of a technology she distrusts because, for months, it was the only employer that answered.

For Gen Z, the appeal of the AI data annotation side hustle is easy to see. The work needs a laptop and an internet connection, the shifts are flexible, and the entry requirements are low compared with most of the jobs that pay above average. It sits next to food delivery and online freelancing in the list of most common side hustles, where a recent LendingTree survey found Gen Z hustling less and earning more; the AI data annotation side hustle is the newest arrival in that lineup. There is no commuting or customer at the door, and no weather to deal with. There is also no workplace to walk away from when a shift ends, and the tasks can be repetitive enough that some workers describe the psychological wear of labeling all day.

What happens next is tied to the models themselves. Each new generation of AI produces more output, and every output still has to be checked, corrected, and re-checked by a human before it can be trusted. The Times reporting suggests the demand curve has not flattened even as the tools improve. For graduates like Amelia, that means the AI data annotation side hustle could also be one of the most durable trades of the current job market, even as others debate whether AI gains could shrink the workweek: teaching machines to understand the world, one labeled image at a time, while deciding how they feel about the technology they are training.