Mary L. Gray: The invisible "ghost" workforce powering our day-to-day lives The technology powering many apps and services seems automatic. But anthropologist Mary L. Gray explains how there are millions of hidden workers behind the screen who are key to making it all work.

Mary L. Gray: The invisible "ghost" workforce powering our day-to-day lives

  • Download
  • <iframe src="" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript


So you may not realize it, but there are companies that go incognito so that they can smoothly and swiftly help you get through the day.

MARY GRAY: So let's imagine you're running late to a meeting or trying to get home and, you know, too far to walk. Maybe you don't have public transit.

ZOMORODI: This is anthropologist Mary Gray.

GRAY: And so you decide to order a Lyft or Uber or some - even a taxi. And after a few seconds, your driver accepts your request for the ride.

ZOMORODI: Let's call him Sam.

GRAY: Unbeknownst to you, Sam shaved off his beard last night. And this only matters because, for a company like Uber, for example, it has all of their drivers verify their IDs with pictures. So the picture verification that a driver like Sam might send in the morning may not match his photo ID that's on record. That's going to set off an automated alarm bill for Uber.


ZOMORODI: Sam doesn't know it, but before he picks you up, halfway across the world, a woman named Ayesha is quickly checking his ID.

GRAY: And in front of Ayesha is going to pop up those two faces - a photo of Sam that's on record and the photo he just took where he doesn't have his beard anymore. And her job is, within seconds - quite literally, as a timer is ticking down - to identify, is this the same person?

ZOMORODI: She squints at her screen with just moments to decide, is the clean-shaven guy really your Uber driver? Ayesha decides, yep, that's Sam.

GRAY: When she clicks yes, it's funneled into Sam's account, and he's none the wiser. He has no idea that his ride could have been rejected.


GRAY: And then you're on your way to the meeting. Sam's behind the wheel. You're both completely unaware that Ayesha has actually facilitated you being able to get back to the office or to get home in time for dinner.

ZOMORODI: This work, this micro-job that Ayesha did, is what Mary describes as ghost work. For the past decade, she's been researching how millions of people like Ayesha make our interactions with companies like Uber, Amazon, Google and Microsoft appear seamless.

GRAY: In many ways, it's the people behind the screen who are doing content moderation, data labeling and a host of other activities that, for the most part, we have no idea are integral to making the internet work.

ZOMORODI: So me - in this scenario, I trust that the app is not going to send a, you know, axe murderer to come pick me up. Why? Why doesn't the app want us to know that there are these humans doing really important work to keep us safe? Why do they keep it under wraps?

GRAY: Well, the hard part is that in most cases, the assumption is, as the consumer, you don't want to be bothered. But there isn't a way to have this kind of what seems frictionless, smooth experience as a consumer without having also a process that verifies who's behind the wheel. There is no computational system that can identify a person with 100% accuracy every single time. And so Ayesha becomes necessary just for a moment. But we're sold the magic that Ayesha doesn't exist at all, that it's just the algorithms doing the work.

ZOMORODI: To be clear, it's not as though everything on these apps - like when you call a car or you get food delivered or whatever the case might be - it's not all invisible humans working behind the scenes.


ZOMORODI: They - it's a - it's this crucial combination - right? - of algorithms, artificial intelligence and humans. Where does the line get drawn?

GRAY: So the dirty little secret of the past 10, 15 years is that there is no way to automatically update information and verify its accuracy. So we've always had people involved in a moment of looking at information and saying, yes, that's the accurate spelling of a book; yes, that is information that is still living on a website. It's not just a dead link. There's some of that work that can be automated, but there's quite a bit that cannot. And so it's a moving target for computer science and engineering to figure out what can you automate.

ZOMORODI: I wonder - you know, it feels like smoke and mirrors a lot of the time. And I guess it feels like ghost work and not telling us that there are humans doing a lot of the things that we take for granted when we use our phone and computers - like, it feels like deception.

GRAY: Yeah. So I think tech companies are pretending to be something they're not because they're kind of faking it until they make it. As long as it seems like we want things done for us through robots or through AI, they'll keep trying. So this is really important because we haven't had the general public aware enough that artificial intelligence cannot solve all the problems. But I think also we are just coming to grips, both the engineering parts of the house and society, with, wow, this is intractably hard to fix.

ZOMORODI: When we come back, more with Mary Gray and about what it takes to be an on-demand online worker. It's the TED Radio Hour from NPR. I'm Manoush Zomorodi. We'll be right back.

ZOMORODI: It's the TED Radio Hour from NPR. I'm Manoush Zomorodi, and on the show today, going incognito. And we were just talking to anthropologist Mary Gray about the mostly invisible workforce that keeps many of our apps and services running.

GRAY: It's the people behind the screen who are doing content moderation, data labeling and a host of other activities that, for the most part, we have no idea are integral to making the internet work.

ZOMORODI: These companies hire people to do short, quick tasks - ghost work. And this workforce, Mary says, is growing quickly.

GRAY: 2016 - there was a research study that was done by McKinsey, and they estimated that in the U.S. and Europe alone, there are around 25 million people who have done some form of this on-demand gig work online. At that rate of growth, if you combined the current trend in contract staffing, temp agency services, that's, like, 60% - six zero - 60% of today's global employment could likely be converted into some form of this kind of on-demand gig work by 2055.

ZOMORODI: Whoa. That's a lot of people.

GRAY: That's not that far out.

ZOMORODI: Yeah. And it's not like these are easy jobs. You have found that people are often paid cents per minute, and they have to work incredibly fast but also be extremely accurate.

GRAY: Yeah, the intense pressure and low pay is fairly common. And that's because this is - I would argue today, it's unregulated work. I mean, this is contract labor that is constantly changing. It's like a steady stream of what's next that you need to vet or classify or sort. And at the same time, it's cognitively draining. I think the irony here - let's take the example of content moderation. When we started this work, I think most people imagined that that was work that would go away once we automated identifying content that should be removed.

ZOMORODI: Right. I mean, that's what Mark Zuckerberg told us, right?

GRAY: I think that's what he believed. I mean, I think if you're a techno-optimist or someone who's just relentlessly committed to fixing something with technology, it's awfully hard to see problems that are perhaps beyond the reach of technology.

ZOMORODI: OK. But we have to say there is a plus side to some of this work, too, right? I mean, it fits in with people's lives. In some cases, they may not be able to work at all if they don't take these tasks.

GRAY: Yeah, it's appealing. And I think it's important to see, why is it appealing? It's a response to what's not working in employment today. We met people who were doing this work because they had no choice. There were no other good job options. They were trying to control their schedules that often had to do with family care. So for many people, it's about controlling their time. The other thing they're trying to control is what they work on. You know, they wanted to be able to pursue interests.

So, for example, Natalie was someone we met - African American woman who was living with her parents in Queens. She was in her late 20s, early 30s. So for her - for Natalie, this was a chance to be able to do work that could generate an income and also balance her interest in music and choreography. So she wanted to balance her - what she was working on that - in a way that would give her room to do what she really wanted to do, which was her art.

ZOMORODI: The theme of our show is incognito, and I was trying to think - I'm like, so who is incognito? Who is in disguise? Is it the ghost workers themselves, that they are disguised by the technology? Or is it the companies who are saying, here's where we are. We're amazing. Look at what we can do, and it's so fast and it barely costs you anything. Use Uber. It's only $5 for you to drive, you know, all the way to Midtown, in my case. Who is doing the disguising? Who is incognito in your mind?

GRAY: This is where it is really important to say the workers are not incognito. Just because we can't see them doesn't mean they aren't carrying out really valuable work. And importantly, they see each other every day online. They connect with each other. I would say right now, it's the tech companies that are incognito, that have - whether with intention or not - put off recognizing - seeing the value of the people who are critical to their contributions, to their services to consumers. To a lesser extent, in some ways, consumers are just asleep at the wheel. And now that we are learning to pay attention to how many people have a role in us having an experience online or being able to have an app whisk us away or bring us food - now that we know that involves even more people than we realized before, are we ready and willing to pay our fair share?

ZOMORODI: So if you're saying that these jobs aren't going anywhere, that they're absolutely needed by these companies, why don't they build this workforce into something that is reliable and high quality?

GRAY: Well, I think we could.


GRAY: Most of the labor laws that we reference as the things that create security, that create safety on the job, they come from the 1930s, literally. Like, we have no good laws in place that have recognized the current relationship we have to technology and how it mediates all of our work lives. We haven't even begun to think about what that means globally and for a globally connected workforce that works around the clock through the internet.

ZOMORODI: That was anthropologist Mary Gray, and she wrote her book "Ghost Work" with computer scientist Siddharth Suri. You can see her full talk at

Copyright © 2022 NPR. All rights reserved. Visit our website terms of use and permissions pages at for further information.

NPR transcripts are created on a rush deadline by an NPR contractor. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.