Can AI Improve Diversity in the Workplace?
The business case for diversity in the workplace is strong.
A McKinsey study shows that companies in the top quartile for gender diversity on executive teams were 21 percent more likely to outperform on profitability and 27 percent more likely to have superior value creation. Companies in the top quartile for ethnic and cultural diversity on executive teams were 33 percent more likely to have industry-leading profitability.
According to Morgan Stanley, the top third of companies, based on their percentage of female employees and other metrics of gender diversity, experienced 2 percent higher average relative returns than other companies in their region. Over a six-year period, companies with more gender diversity enjoyed a one-year return on equity that was 1.1 percent better than companies with low female representation.
According to the Harvard Business Review, businesses that have greater levels of diversity and inclusion are 45 percent more likely to grow market share year over year and 70 percent likelier to capture new markets.
Hiring employees who bring diverse perspectives and experiences to the table can help a business better understand how to create innovative products and services to meet buyers’ needs, both as they enter new markets and grow existing ones.
Companies are also more conscious about building a diverse culture. Seventy-five percent of respondents to a Boston Consulting Group survey said that diversity is gaining momentum in their organizations.
But it can be challenging to create a diverse talent pool and also to retain these employees. Unconscious bias is a culprit here, both in the wording and tone of job descriptions and in judging candidates’ value to the company and fitness for a job.
Artificial Intelligence Can Reduce Unconscious Bias
Today, businesses are exploring whether artificial intelligence technology can help reduce bias in hiring and employee valuation. Gartner has found that 23 percent of organizations already piloting or using AI have been doing so in the HR and recruiting domain.
Diversity issues in employment start with the way that jobs are described in ads. The Zillow Group wanted to learn why women and minorities weren’t applying for jobs at the company in larger numbers. It used AI to help it to flag language that might be considered insensitive or off-putting to minority groups. It found that its postings were written in a slightly “masculine” tone and suggested wording that added a more “feminine” tone to the postings, such as making them more personal by using the word “you.” In 2018, Zillow saw a 12 percent increase in female applicants compared to the previous year.
AI also can support “blind” applicant screening — so that the person in charge of advancing a candidate to the next level won’t be influenced by their demographic identity — through redacting the parts of a resume that provide identifying information. Anonymized profiles mean that candidates can be evaluated on their hard and soft skills.
Companies also grapple with the problem of the pay gap that exists between male and female workers and among ethnic groups. Talented employees who feel that their sex or ethnic origin is negatively influencing their ability to advance in a company aren’t likely to stay there very long. AI and machine learning technology can help here too, using HR and payroll information to produce data visualizations of insights to help companies understand pay status across departments and even business locations so they can address equal-pay issues.
READ: What kind of tech do companies with happy millennial workers have?
AI Algorithms Can Sometimes Display Their Own Biases
Still, companies should be cautious as they begin to embrace AI to help create more diversity in their organizations. Algorithms can be biased too.
For example, one large employer created a machine learning recruiting engine that turned out not to be rating candidates for technical jobs in a gender-neutral way because of the data it used to derive its computer models. The software vetted applications based on resumes the company received over a 10-year period, but its model was learning from incomplete data that compromised the results: Most of the resumes it trained on came from men, since that’s who typically applied for technical jobs in the first place.
Tech companies are working to minimize such issues. AI will have an increasing footprint in all aspects of the hiring process, and experts say that by 2025, AI will be significantly integrated into recruiting.
