Join getAbstract to access the summary!

Ghost Work

Join getAbstract to access the summary!

Ghost Work

How to Stop Silicon Valley from Building a New Global Underclass

Houghton Mifflin Harcourt,

15 min read
8 take-aways
Audio & text

What's inside?

The global, invisible, on-demand labor market reveals a lot about the future of employment.

Editorial Rating



  • Eye Opening
  • Bold
  • Hot Topic


Researchers Mary L. Gray and Siddharth Suri detail an invisible segment of the labor market: people who perform “ghost work” or “on-demand” labor in the technology industry that algorithms and AI alone can’t do. The authors reveal the flaws in how society undervalues gig work and part-time jobs. They draw a historical line from marginalized workers doing piecework in sweatshops to today’s gig workers, an important analogy since on-demand work is likely the future of employment. Gray and Suri advocate legal and societal changes to make gig work viable for workers and businesses.


Behind the seamless performance of software applications lurks an invisible army of people doing “ghost work.”

The countless tasks that behind-the-scenes workers perform in the United States and worldwide is called “ghost work” or “on-demand work.” The people who do ghost work fill the gaps between what AI and algorithms can accomplish and what the end consumer expects. They apply ingenuity to problems that software cannot yet solve.

Software companies use on-demand workers to train AI applications these firms hope will replace much of their human workforce. As applications become more powerful, human labor makes them competent and responsive. Ghost work may consist of “micro-tasks,” such as the Human Intelligence Tasks (HITs) on Amazon’s Mechanical Turk platform, which comprise actions as small as tagging an image. Workers on Upwork and Fiverr facilitate “macro-tasks,” like copyediting or website design. Full time staff members at hiring companies often manage on-demand workers.

Ghost work is the latest incarnation of part-time, marginalized, devalued labor.

Using on-demand, ...

About the Authors

Mary L. Gray is a senior principal researcher at Microsoft, a fellow at Harvard’s Berkman Klein Center for Internet and Society, and a faculty member at Indiana University’s School of Informatics, Computing and Engineering. Senior principal researcher Siddharth Suri was a founding member of Microsoft Research, New York City, before joining the Human-Centered AI group in Microsoft Research, Redmond, Washington.

Comment on this summary