AI talent is scarce and companies cannot afford to overlook half of the population through systemic and cultural inequities.

“We've made progress, but we still aren't where we want to be,” said Aalya Dhawan, director of global communication and content at the nonprofit Women in AI. The organization will be presenting at next week's AI Summit London, on June 14 to 15.

Only 4% of respondents said that their organizations had achieved 50-50 parity in employment between men and women, according to the 2023 survey "Advancing Gender Equality in Data, Analytics and AI," conducted by Omdia, AI Business and Women in AI.

Moreover, 39% said their companies would take three to five years to achieve parity while 54% believe it would take 5 years or more.

“Bias may sometimes emanate from both the top down and the company culture, which is why it's important to support more efforts and dialogues with women,” Dhawan said. “We want men to actively support women as well, so it's not just about women helping women. It's a deeply ingrained social system that we must address now more than ever before.”

Tellingly, 47% of respondents said they didn’t know what their companies were doing to achieve equity employment between male and female workers. Of the 101 global respondents, two-thirds were CEOs, company founders, C-level executives or middle managers, those most likely to know their diversity situation. All work in the AI field.

The good news is there is awareness of this shortcoming, with 57% indicating that their organizations have made or are taking steps to improve gender diversity, and 52% said diversity is an organizational priority.

However, AI companies still need a more granular understanding of diversity and equity challenges in order to progress further.

Stay updated. Subscribe to the AI Business newsletter

Gender cultural barriers may be embedded in organizations that limit women’s advances, and managers and recruiters may not even be aware of such biases, according to the study.

It starts with the recruiting process, where many HR systems and practices show gender bias. For example, a Proceedings of the National Academy of Sciences (PNAS) study concluded that both male and female recruiters associated women with lower math and science skills than men.

Our survey recommends four steps to address the gender gap in AI:

1. Engage more women in STEM programs early

STEM (science, technology, engineering and mathematics) is an education initiative designed to encourage students from elementary through college to enter the fields of technology, engineering, mathematics and science. STEM programs operate in Australia, China, France, South Korea, Taiwan, the U.K. and the U.S.

In assessing STEM progress, the American Association of University Women reported that men vastly outnumbered women majoring in college STEM fields. It cited causative factors such as gender stereotypes (STEM fields are generally viewed as masculine), male-dominated STEM cultures, few female role models and early socialization of girls that generated anxiety and a lack of confidence in mathematics skills.

2. Revisit corporate recruitment practices

AI companies can improve their ability to attract top talent by screening for bias in the rulesets and algorithms used by their automated job application screening systems. If biased criteria are found, they can be amended. Companies can also work with colleges and universities by providing internships that students use for course credit, and by hiring the best and brightest of these interns for permanent positions — including women.

3. Make gender equity a priority in your work environment

In the report, 69% of respondents felt that an improved workplace culture correlates to a more equitable workplace. Tellingly, 73% said men outnumbered women in their departments.

For AI companies, hiring and attaining equal numbers of male and female employees would be the easiest metric for closing the gender gap. Other equity indicators such as equal access to promotions and earnings opportunities are more elusive.

4. Provide advocacy that advances gender equity in educational, corporate and government settings

In our survey, 53% of respondents said they were mentoring at least one person and 41% said they were mentoring more than one person. This is a positive start and can be expanded to providing mentoring, guidance, guest lectures and role models to young girls and women in school and college classrooms.

AI companies can help by offering maternity/paternity leave, flexible hours and affordable childcare. They also can work with government agencies to boost female participation in STEM programs, among other initiatives.