When people talk about AI and jobs, the story usually goes one way: AI automates and humans lose out.

The reality is far more nuanced. Behind these advanced models are the people who teach them how the world works, correct its mistakes, and decide what “good outputs” really look like. This work is called data annotation, and until now, its economic impact has been largely unknown.

To better understand the industry, Scale commissioned Oxford Economics to conduct the first comprehensive economic study of the broader US data annotation industry, The Economic Impact of the Data Annotation Industry. The findings confirm what we have seen up close for years. Data annotation is not just a hidden step in the AI pipeline. It is a growing part of the US economy, and it is creating flexible earning opportunities for tens of thousands of people.

Scale serves a variety of customers across the full AI lifecycle, constructing the foundation of AI systems, building enterprise and government AI applications, improving AI training data, and leading independent model evaluation. To support these products, part of our business involves working with expert contributors. Contributors are independent specialists that help create and curate the high-quality data that is needed to produce high-quality AI.

But who are the human experts annotating, training, and evaluating this data, and what impact do they have on the economy?

An Emerging Industry Powering the Future of AI

Oxford Economics estimates that the US data annotation industry contributed $5.7 billion to US GDP in 2024, with total impact projected to grow to $19.2 billion by 2030.

That impact comes from more than just a handful of companies in the AI stack. In 2024, data annotation work in the US:

Supported nearly 200,000 flexible earning opportunities
Sustained about 9,000 full-time employee jobs directly
Supported another 25,000 full-time employee jobs through suppliers and local spending
Generated over $1.2 billion in federal, state, and local tax revenue

In other words, data annotation is already on par with established manufacturing sectors and is growing far faster.

Who is doing this work?

The people embedding wisdom into AI come from all kinds of backgrounds, are often experts in their respective fields, and take on this work for a range of reasons. The results of Oxford’s study confirm what we see every day: these contributors are highly skilled people, dedicated to their professions, to their families, and to their communities.

The study brings the people behind those numbers into focus. They’re individuals with real-world expertise who help build AI.

84% of respondents hold at least a bachelor’s degree, roughly double the share in the broader US workforce.
About 65% are under 44, meaning the workforce skews younger than the national average.
94% are also engaged in other work, education, or caregiving responsibilities; meaning data annotation provides supplemental income.

They include software engineers, researchers, teachers, health workers, creatives, caregivers, and students who bring subject-matter expertise into AI systems. Many already have demanding jobs or family responsibilities and use data annotation as a way to earn additional income on their own terms.

Some, like Fred Nau, a high school chemistry and physics teacher from Miami, Florida, used the income he earned to save up for a ring for his fiancée and for the down payment on a place of his own.

Fred’s story, like many of our contributors, is a fascinating one. Over the next few weeks, we are going to continue sharing the stories of contributors like Fred, giving more color to the people and places behind the AI industry.

Why flexible work matters.

For most contributors, data annotation provides an earnings opportunity that fits around their lives.

94% say earning additional income drew them to the work.
91% say flexibility is a core reason they do it.
87% use this supplemental income to offset lifestyle costs
Nearly half use it to build savings, and about a third use it to support family members.

This flexibility is driven by the short-term, project-based structure of Outlier tasks, giving contributors control over when and how much they work. Two in ten contributors in the study report living with a disability. For them, data annotation is one of the ways to earn income without sacrificing their health or stability. As respondents put it, this kind of work is appealing:

“because I can work or not work based on how my symptoms are,”
“I can take time off for frequent doctor visits or bad days without losing my job,”
“it lets me use my skills meaningfully while keeping my dignity despite my disability.”

Students use their earnings to pay tuition or save for the future. Caregivers fit projects around childcare, eldercare, or other health needs. Small business owners smooth out income between projects or invest in their own ventures.

The future of the industry, and the people within it

Many annotators are already domain experts. They are software engineers, researchers, teachers, and specialists who are deepening their understanding of how AI systems behave and how they interact with their chosen profession.

Still, many contributors are using this work as a springboard for their future goals and career aspirations.

Almost three-quarters of our annotators hope to use their AI skills in future AI or tech roles
69% believe their AI skills will be transferable across industries
64% believe their AI skills from annotator work offer good opportunities for future career progression.

Over the next few weeks, we will continue to share the stories of our contributors, talking to a mom and educator from Alabama, a dad looking for flexible earnings opportunities in rural Louisiana, and a former NASA engineer who lives in Virginia, just to name a few.

We look forward to sharing these stories and highlighting the people who contribute to the AI ecosystem.