Diversity in the American workforce is dismal. According to the US Equal Employment Opportunity Commission, in 2014, only 1.9% of college-educated workers were black. Among Hispanics, that number was 4.4%.
Though there are several historical reasons for these imbalances—in tech jobs, for example, the low number of women majoring in STEM subjects is particularly discouraging—the current perpetuation of the problem can be traced partly back to the hiring process. More specifically, blame can be laid on the way some employers phrase job descriptions, which can dissuade women and minorities from applying.
To eradicate these systemic biases, a few startups specializing in machine learning and artificial intelligence think they can help—with linguistics.
Textio has helped companies increase female job applicants by 23% compared to previous hiring rounds. (The software also draws in 25% more candidates overall who are qualified to make it to the interview round.) Kieran Snyder, co-founder of Textio, says they are able to get these results by eliminating phrases in job ads such as “coding ninja” or “fast-paced work environment,” which tend to attract mostly white males. To replace these terms, the tool suggests inclusive phrases such as “collaborative” that appeal to a broader range of potential employees.
Textio is built on a dataset of over 240 million job posts that are being cross-referenced with real-world outcomes. “As customers use Textio to write their content, they contribute their anonymized data—published job posts, applicant stats, and hiring outcomes—to the collective dataset,” Snyder says. As the user base increases, the tool will become progressively smarter, too.
Talent Sonar, another company that aims to improve diversity through technology, uses thousands of surveys to figure out prevalent associations with words and recommend alternatives. “From birth our brains are forming neural pathways that relate every word to the various images that are associated with that word,” says Laura Mather, co-founder of Talent Sonar. For example, the word “competitive” is often associated with a picture of a sports team (usually male, often white), so the brain makes a connection between the word and the picture. This means that when non-male/white candidates see that word in a job description, they will subconsciously not imagine themselves in the image the company is presenting, and be less likely to apply.
By testing words to see how they influence people’s responses to job descriptions, Talent Sonar is trying to expose dominant neural pathways. “What we’ve found is that by balancing the words that are often associated with white males (“competitive”) with words that are associated with other demographics (“collaborative”), you can engage neural pathways associated with lots of types of images,” Mather says. “This makes a larger number of people feel welcomed by a specific job description.”
The work doesn’t end with job ads though. Talent Sonar also helps create a blind resume-evaluation process that prevents reviewers from getting distracted by gender or ethnicity. In a world where someone with a white-sounding name is more likely to be invited for an interview than someone with a name typically associated with minorities, this step plays a vital role in getting companies to hire employees based on merit. Similarly, the software formulates a series of questions for the interview phase that encourage the HR team to focus on values and objectives as opposed to being swayed by race or gender.
Although both companies are still young, having been founded less than five years ago, they predict their brand of augmented writing will influence other areas of business in the future as well. “Our customers already use Textio for more than just job listings,” Snyder says. “They are editing marketing content, sales pitches, and much more.” The move toward targeted, non-biased writing in all professional realms represents a real opportunity to make communication more inclusive. It won’t solve all the imbalances in the workplace, but it’s a good start.