A world in which humans are monitored and overseen by machines? Amazon, the e-commerce giant, has been the subject of frequent reports in recent years of overstress in its warehouses, with employees struggling to process hundreds of boxes of goods an hour — and quickly being fired if they don’t move fast enough.

And if it’s artificial intelligence that’s monitoring your productivity, it’s even grimmer. A recent document obtained by The Verge suggests that far more workers are being laid off for lack of productivity than outsiders realize. Between August 2017 and September 2018, a single Amazon warehouse fired 300 people, and crucially: most of those dismissals were automated, according to the letter.

The leaked document is a correspondence between Amazon lawyers and the NATIONAL Labor Relations Board regarding a case: cdn.vox-cdn.com/uploads/cho…

Simply put, Amazon’s AI systems can track each employee’s productivity and automatically generate warnings or firing orders, without any involvement of a human supervisor in the decision-making process. Although Amazon officials say managers have more authority than AI, the company has not been more specific.

This isn’t the first time a human has been “fired by a computer.” Drivers for Uber and other ride-sharing services have complained before about how AI algorithms are managed, and these stories may become more common as automation advances.

Part of the file that was exposed.

Amazon Fulfillment Center is the company’s engine — employees track, pack, sort and ship each ordered package in a giant warehouse before it’s sent to sellers.

Pressure on Amazon’s delivery services has been growing, and there have been frequent comments in recent years that these workers are facing tough conditions: employees are forced to increase their “work speed”, with some packing hundreds of boxes an hour, or losing their jobs if they fail to do so. Stacy Mitchell, co-director of the Local Independence Association (ILSR) and a prominent Critic of Amazon, says: “Logistics workers in this situation are always backed up by someone who can do exactly what they do.”

Documents obtained by The Verge show that firing employees for productivity is far more common than laypeople think. In a signed letter last year, a lawyer representing Amazon said the company had fired “hundreds” of employees at one warehouse alone between August 2017 and September 2018 for failing to meet productivity targets. During that time, about 300 full-time employees were laid off for inefficiency, according to a company spokesman.

That number is representative of most warehouse workers: The Fulfillment Center in Baltimore now has about 2,500 full-time employees, a spokeswoman says. Assuming the firing rate stays the same, that means Amazon is firing more than a tenth of its employees each year for productivity alone. For North America as a whole, the numbers are even more striking. Amazon operates more than 75 fulfillment centers with more than 125,000 full-time employees, meaning the company puts thousands of employees out of work each year because it can’t move packages fast enough.

The documents also show a highly automated tracking and firing process. “Amazon’s systems are able to track the productivity of each employee and automatically generate warnings or dismissal orders about work quality or efficiency without the involvement of a human supervisor,” according to the signed letter.

“Warehouse employees are monitored and supervised by robots.”

Critics see the system as a machine that only sees numbers, not people. “What we hear from employees all the time is that they’re being treated like robots at work because they’re being monitored and supervised by these automated systems,” Mitchell says.

The system can even track “idle time jobs,” which the company abbreviates to TOT. If an employee fails to scan a package for an extended period of time, the system automatically generates warning instructions and eventually fires the employee. Some warehouse workers say they forgo bathroom breaks in order to be as productive as expected.

Amazon says retraining is part of its process to bring employees up to standard, and that it will change the firing rate only when more than 75% of its warehouse staff meet its goals. According to the company, the bottom 5 percent of employees receive retraining programs. The grievance system is also part of the dismissal process.

An Amazon spokesperson said: “Between August 2017 and September 2018, approximately 300 employees at the Baltimore warehouse left for productivity reasons. But overall, layoffs at the Baltimore warehouse and other warehouses in North America have decreased over the past two years.” Amazon did not give details of its current firing rate.

Amazon provided the data in a labor dispute in which a former employee at the Baltimore warehouse filed a complaint with the National Labor Relations Board saying they were fired for performing legally protected operations. In a letter to the committee, Amazon responded that the employee had been fired for failing to meet productivity standards, which is a common occurrence. To bolster its case, Amazon also provided a list of employees fired from its Baltimore warehouse, labeled BWI2. The Verge obtained The letter and related documents under The FREEDOM of Information Act.

“Amazon has a history of firing fulfillment center employees who repeatedly fail to meet standard productivity,” the company’s lawyers said in the letter. So, Amazon fired the employee. “Amazon has also fired hundreds of other employees for the same reason, unrelated to any alleged protected activity,” the lawyer wrote in the letter. The former employee’s claims were eventually dismissed.

“Employees must process orders carefully and efficiently.”

Although the names of the fired employees submitted by the company are bludgeoned out of the document, as many as 900 entries also identify each employee’s supervisor and the reason for the firing. We can see that all the employees on the list were fired for “productivity,” or a category of violations known as “productivity bias,” which can be explained as a series of inefficiencies. Amazon says the results so far show problems that are too broad — including other performance issues — and that it is working with its board to fix them.

The disclosure also sheds new light on amazon’s strict internal standards. “Employees must be careful and efficient in handling each order,” the letter said, and “productivity metrics” were developed to ensure efficiency. Amazon says the metrics are objective and are based on customer needs, regions and other contexts.

Amazon workers have sometimes rebelled against the company’s productivity requirements. Last year, East African migrant workers at a Minnesota warehouse organized a protest against the company, saying they did not get enough rest time, including prayer time.

In response, Amazon continued to tout the company’s care for its employees, pointing to their hourly payback and numerous benefits like maternity leave. But the document makes it clear that for people who don’t meet productivity standards, they don’t get those benefits at all.

While it may seem reasonable to dismiss people if they fail to perform, there will always be a sense of ruthlessness about automated management using methods such as artificial intelligence, since humans can’t always produce at the same rate as machines. As Amazon’s business continues to expand, it doesn’t look like machines will be replaced any time soon.

Reference content:

  • Futurism.com/amazon-ai-f…

  • www.theverge.com/2019/4/25/1…

  • www.businessinsider.com/amazon-syst…