Skip to navigationSkip to content
Reuters/Stephen Lam
A little side cash never hurt nobody.
MINTING MONEY

Low-paying crowdsourced jobs are attracting a lot of young, college-educated Americans

Ananya Bhattacharya
By Ananya Bhattacharya

Tech reporter

From our Obsession

Machines with Brains

AI is upending companies, industries, and humanity.

In 2005, Amazon created the site Mechanical Turk to give academicians and companies a platform to find people to complete quick, menial tasks than computers cannot handle. These were dubbed Human Intelligence Tasks (HITs).

Since then, the site has registered over 500,000 workers worldwide, according to a 2015 World Bank report. Who are they? A new Pew Research study surveyed 3, 370 US “Turkers” between Feb. 9 and Feb. 25 and found that young, college graduates are swooping up these meager tasks.

Over 88% of the Turkers were under 50 and half had a college degree. This was a surprise to Pew.  ”You would expect a lower paying job like this would have less-educated people but here, despite relatively low wages, you see a number of educated people doing it,” senior researcher and author of the report, Paul Hitlin, said in an interview. “People can work when they want and where they want and that creates a lot of value for them.”

People can work when they want and where they want

More than half, or 52%, of the respondents reported that they earned less than $5 per hour. Hitlin said that minimum wage laws don’t apply because workers get paid based on what tasks they choose to do and how fast they work. Only a quarter of the people on the site said that most or all of their income came from Mechanical Turk tasks. Most people just use it as an additional stream of income. 

About 61% of the tasks that can be completed in under a few minutes paid less than 10 cents each, researchers found.  Scholars who use the site do so for one-off or short-term projects.

For companies, Turkers are cheaper than machines. Some businesses have made the site part of their ongoing business models. For instance, identifying a receipt and logging it may seem like an easy job for a machine, but computers can fail at this task: The receipt could be crumpled, faded, not aligned well, or it may be from a store the machine has never seen before.

Humans can recognize these things almost instantly.

Subscribe to the Daily Brief, our morning email with news and insights you need to understand our changing world.

By providing your email, you agree to the Quartz Privacy Policy.