Audible subscribers can listen to all episodes of business wars ad-free right now.

Join Audible today by downloading the Audible app.

It's 2025 in Nairobi, Kenya.

A data labeler clocks in for his shift and lets out a heavy sigh.

He does not want to be here this morning.

He walks into a large room lined with rows of workers tapping away at computer terminals.

He logs in, opens a program, and starts to watch a woman get undressed. She seems totally unaware that anyone might be viewing her, but this data worker can see everything because the video is being streamed live from a pair of meta-glasses sitting on her nightstand, a continent away.

The Kenyan man works for a data labeling company called Sama, a corporation that supplies the human labor behind the so-called autonomous machine

learning at the heart of AI. Every day, in 10-hour shifts, his job is to review unfiltered video

content and label what he sees by drawing bounding boxes around the images on the screen. He draws one box around a cluster

of pixels and labels it a lamp. He draws another box and labels that a bookcase. He's training

the AI to recognize everyday items so the algorithm can get better at doing it in the future.

The labeling work is low-paid and repetitive, but that isn't the worst part, because this man

doesn't just see lamps and bookcases. He often finds himself reviewing sensitive or graphic

video content, like the woman undressing.

He sees strangers' credit card numbers.

He sees people having sex.

He even sees crimes being committed.

All of it captured in first-person video,

streamed by the people who have no idea they're broadcasting their lives to the cloud, much less that it's

being reviewed by a human. And more and more of this footage now seems to be coming from

one device, meta's rayband glasses. You might know them as meta smart glasses or meta-AI

...