A Cautionary Tale: Amazon’s Experiment with AI in Hiring


While the search for true human-level computing power is still elusive, the targeted use of artificial intelligence is currently transforming individual lives and entire industries. From intelligent sourcing and automatic ordering optimizing supply chains for manufacturers to AI assistants routing incoming telephone calls and enhancing customer service, the examples are all around us.

These specific uses for artificial intelligence are already well known, at least within the industries where their impacts are being felt. What is less known, but potentially far more important, is the conclusions AI will lead us to, especially in areas like hiring.


Amazon as an AI Proving Ground

The use of automation to screen resumes and artificial intelligence to guide hiring decisions has begun to be implemented at some companies. Arguably the most high profile examples of artificial intelligence guiding the hiring process happened at Amazon and this experience serves as a cautionary tale of its current limitations.

At first blush, it would appear that Amazon is the perfect proving ground for an algorithm that automates the hiring process and facilitates the sorting of resumes. The online retail giant is notoriously choosy about the candidates it hires, and the company can receive thousands of resumes for every job opening.

Given the sheer volume of applicants, some type of automation is virtually essential for Amazon. So when the company began experimenting with an artificially intelligent hiring assistant, hopes ran high. Unfortunately, the reality was less than spectacular, and the company announced it was discontinuing the project.


Discriminating Against Women

Almost from the start, there were issues with the Amazon AI hiring tool. Some of those issues arose right away and were very obvious, while others initially flew under the radar, potentially making them far more insidious and much more dangerous. One of the biggest problems with the Amazon experiment was the way it skewed hiring decisions in favor of male applicants. Tech industry watchers and government officials have long decried the male-dominated nature of work at high tech firms like Amazon, but the artificial intelligence tool appears to have made the hiring process even more imbalanced.

The team behind the artificial intelligence recruiting tool Amazon used had been hard at work since 2014, building in real-world experiences and the unique recruiting challenges faced by other high tech companies. Even so, the bias shown by the tool eventually proved problematic, and the company claims it is no longer relying on AI to recruit new workers for its various warehouses and offices.


Talent Optimization as a Human Approach

While some hiring tools are still in experimental stage, smart options currently exist to help companies build a world-class workforce. Talent Optimization, for example, provides a framework for hiring that leaves the human element intact. Through aligning an organization’s business strategy with its current workforce and future hires, this data and analytics tool helps companies make smart hiring decisions without gender bias.


AI Limitations

In the short term, no one truly knows what the future of artificial intelligence will be when it comes to making hiring decisions. In some cases, the recent Amazon experience should serve as a cautionary tale, uncovering the current limitations of a technology that, for all its potential, is still in its infancy.



About the Author: Shelley D. Smith is a best-selling author and Talent Optimization Advisor PI Midlantic. Her experience over the past 35+ years has earned her a reputation as the Creator of the Culture Inquiry in businesses all along the east coast, and beyond.