Remember when Amazon scrapped its artificial intelligence program for recruiting because it was biased against women? That wasn't the end of computer decision making for Amazon. The Verge uncovered documents from Amazon showing that they use computers to warn and fire warehouse workers based on productivity.

While I'll argue all day that it's critical for companies to put my human into their human resources, I'm okay with this type of computer led termination scheme. Here's why.

Productivity is objective in a warehouse environment

One thing Amazon doesn't lack is customers, so there are always orders to be filled. When you work packing boxes, there's a clear measure of productivity that can be tracked. Technology allows that to be tracked automatically and in real time. When you're not living up to the productivity requirements, it's time for a warning and ultimately a termination.

It's very different than trying to measure productivity in, say marketing, as it's difficult to track idea creation and brainstorming sessions. In this situation, there are clear indicators of when someone should be notified.

Documents obtained by The Verge state:

"Amazon's system tracks the rates of each individual associate's productivity and automatically generates any warnings or terminations regarding quality or productivity without input from supervisors." 

This is something that should be tracked automatically. It removes all bias--the supervisor's best friend will be held to the same standards as her worst enemy. Now, supervisors can override such terminations, but, presumably, an override will come with inquiries as to the reason. That's how it should be.

It makes the supervisor's job easier

Firing people is hard. Putting people on performance improvement plans is easier, but still quite difficult. This helps take the emotion out of the situation. Objectivity helps with that. And because of the automated nature, you can eliminate the hemming and hawing that often comes with the decision to speak to an employee about her performance.

It's not done in a vacuum

Amazon doesn't just set goals and walk away. They retrain when people struggle. The adjust goals so that 75 percent of the organization is meeting them. If 75 percent are meeting, you can say it's probably attainable for just about everyone. The supervisor override can help take care of special situations.

But, overall, it helps prevent bias and holding one group of people to on standard and a different group to a higher standard. Computers don't care about race, gender, or age (unless you tell them to).

What about ADA?

The Americans with Disabilities Act requires that companies provide reasonable accommodations for disabled employees. The employee, with the provided accommodation, still needs to perform the core functions of the job. In this case, packing boxes at a certain level of productivity is the core function. As long as they are providing accommodations when needed, this wouldn't violate any disability protections.

And the humans?

My preference is that each employee would be notified by their supervisor, but I have no problem with computers determining who fails to meet productivity goals. It's not a system that would work in every type of environment, but it works in this one.

Amazon has long had a reputation of being difficult to work for, and this just adds to that reputation. As I've said before, this is not a job or a company that I would want to work for. But, enough people do want to work for Amazon that they can be confident in how they decide to hire and fire.

Published on: Apr 28, 2019
The opinions expressed here by Inc.com columnists are their own, not those of Inc.com.