Your Tech Is As Biased As You Are  

Unconscious bias not only affects hiring decisions — it also affects hiring systems, data, and processes. For example, in 2014, Amazon engineers built an AI-powered hiring tool to auto-select the best applicants from a resume pool. The goal was to automate the resume-screening process, saving recruiters valuable time finding and qualifying top talent.  

But a year into the project, the engineers discovered a problem: The AI favored male candidates, particularly for technical positions like software engineering. How did this happen?

Two reasons: First, the resumes submitted for technical positions used to train the AI had mainly come from men. So the technology learned to downgrade references to “women’s” clubs and activities in female applicants’ backgrounds. 

Second, the program learned words and phrases more often used by men to write their resumes. Once the engineers discovered the bias, Amazon scrapped the tool.

Commentators often use this story as an example of “why technology is scary” for recruiting and hiring. But that’s not the lesson here at all. The lesson is that, sometimes, tactics used to improve hiring — from adding new technology to training — could have unintended consequences if we lack an awareness of our unconscious biases that get baked into hiring systems, data, and processes. No matter how good our technology or systems are, failing to address unconscious bias can keep us from achieving our diversity goals. 

This is an especially important point as more TA departments are using technology to reduce unconscious bias in their hiring processes, such as systems that mask names and images from a candidate’s application. However, remember that no single technology is a silver bullet. In addition to using the right tools, TA can make hiring less susceptible to unconscious bias if they make holistic changes the right way. 

Based on my work implementing technology to help achieve D&I goals for some of the fastest-growing tech companies, here are three HR changes that can mitigate unconscious bias to help build diverse teams.

Change Your Mindset

Thankfully, more companies use D&I efforts to reduce discriminatory thinking and practices in their organizations. Still, even the best tools and programs can be ineffective if employees, from executives to line leaders, don’t see unconscious bias as a problem. Unconscious bias affects us all. 

To address this, everyone needs to understand how unconscious bias can affect their perceptions and decisions when choosing whom to hire or promote. Therefore, effective D&I programs help change leaders’ mindsets from thinking of unconscious bias as a problem “out there” to a problem “in here.” In other words, instead of dismissing unconscious bias as overt discrimination by “other people” and “other teams,” efforts to improve D&I should encourage and equip each person in the organization to look for sources of bias in the decisions they make.   

For example, a company may start a D&I initiative to increase their recruiting pipeline’s diversity by building closer relationships with historically Black colleges or women’s organizations. The company may succeed at adding more diverse candidates to the top of the recruiting funnel but fail to examine how unconscious bias can affect the steps that follow. That is, the organization might attract diverse candidates through a diversity-focused recruitment marketing program but deter candidates from applying due to how job descriptions are worded. 

Also remember that unconscious bias can affect decision-making whenever human judgment is involved. Therefore, D&I initiatives should improve metrics like pipeline diversity and challenge the status-quo rationale for decisions such as how minimum qualifications are set or how interviewers are trained.

Companies can start small when building awareness of unconscious bias as a problem. For instance, a simple first step could be to encourage managers to take and discuss results from the Project Implicit assessment produced by Harvard University.  

Rethink Training

Yesterday’s bias or discrimination training typically used video vignettes showing discriminatory or inappropriate behaviors on the job and how employees should avoid or confront them. Characters who were obviously unconcerned with their biases and flaunted them openly acted out these recognizably bad-behavior scenarios. 

While it’s a good idea to train employees on recognizing and avoiding overt discrimination, such traditional training programs can miss harder-to-detect unconscious biases and are often used as a Band-Aid solutions to increase diversity. Instead, to improve diversity efforts, training should accomplish two things: 

  1. Build employee awareness about unconscious bias (specifically that it exists and how it manifests in our perceptions and decisions)
  2. Offer a starting point for creating equitable practices across the employee lifecycle

Here’s an example of how unconscious bias training helps bring employee awareness that can then percolate into day-to-day hiring practices: After receiving company-wide unconscious bias training, recruiters and hiring managers decide to address how interviewers often engage in hapless or inappropriate small talk. Although this looks like rapport building to most interviewers, unstructured interviews easily let unconscious affinity (or similar-to-me) bias into candidate evaluations. 

Article Continues Below

As a result, the company builds a structured interview guide that every interviewer must follow. Standardized questions and criteria help evaluate candidates on the skills that matter versus other (potentially biased) factors. Armed with a new hiring tool and a greater awareness of bias, the company also develops a program to train interviewers on how to stick to the structured interview process. Greater awareness company-wide, new hiring tools, and practical training on using those new tools help protect against the pernicious ways bias affects human judgment and perception.

Expand and Enforce Measurement

It’s also important to measure post-hire outcomes to meaningfully and sustainably impact D&I. But what does that look like?

Let’s say a company wants to know why, despite hiring more people of color in a calendar year, there have not been overall shifts in employee diversity. First, it’s wise to set the right benchmarks and align them with the company’s strategic goals — benchmarks that are good for one company may not be suitable for another.

However, it’s common for organizations to track recruiting diversity but not the hiring outcomes or retention rates of diverse hires. While figuring out which goals are best for your company, it’s vital to track diversity from pre-hire through to post-hire — in other words, diversity metrics of job offers, acceptance rates, and long-term retention.

Work with your analytics department to analyze recruitment data consistently, and then compare it to workforce data to see trends over time. Without a culture of inclusion, turnover for typically marginalized employees will remain high.

Select the Right Technology

Addressing a complicated issue like unconscious bias in the hiring process can understandably feel daunting! But removing bias from the hiring process is like deep-cleaning your house: it’s easiest when you tackle it “room-by-room” instead of all at once. Similarly, make changes to one part of the hiring process and then another, using technology to reduce or remove the unconscious bias that comes with human judgment-made decisions. 

When selecting the right technology, you don’t need a big budget. Your real challenge will lie in finding the best vendor for you, which is impossible to do based solely on vendors’ websites. Instead, request strong case or empirical studies and consider any lack of this evidence as a red flag. 

Once you have a vendor shortlist, get to know vendors to make sure they will work with you in a consultative way. Don’t accept a one-size-fits-all solution out of the box. Instead, a great vendor will understand what you’re trying to achieve, specifically with D&I.

Ultimately, many companies want to achieve significant results when it comes to D&I but are disappointed with their progress thus far. Don’t be disheartened if you’re one of them! Like Amazon learning from their mistake, it’s important to recognize that this is a process. The push toward progress takes time and effort, but with the right mindset, training, and tech, your company can mitigate bias and improve D&I.

Neil Morelli, Ph.D, is the chief industrial-organizational psychologist at Codility, a pre-hire assessment platform for software engineering talent. Neil focuses on scientifically validating test content, consulting clients on hiring best practices, and providing data analyses on hiring fairness. He has spent more than 10 years in technology-enabled talent acquisition, helping Fortune 500 and venture-backed tech companies create effective and inclusive hiring practices.

Topics