Diversity within your AI team can reduce bias

NEW YORK -- As the AI field becomes increasingly focused on ethical AI and bias in AI algorithms, attention to bringing more demographic diversity to employment in AI is also becoming prominent.

Women make up a little more than a quarter of workers in AI roles worldwide, according to the World Economic Forum.

Asha Saxena, founder and CEO of Women Leaders in Data and AI, is working to bring that number up. Her organization mentors women currently in AI and data leadership roles to help them stay there.

Without diversity in the teams and mindsets that create AI products, most products will not yield desired results, according to Saxena.

In this Q&A at the AI Summit New York on Dec. 8, Saxena, who founded the group in 2020, describes the relationship between diversity, data and AI and why it's not enough for organizations to hire women workers -- they must also put them in leadership positions.

What is the relationship between diversity, data and AI?

Asha Saxena: Diversity is critical to the outcome of anything you do in data or AI. You won't get the right result if you don't have enough data. So, you want to ensure that there is a diverse data set to bring you the right results.

Somebody I was coaching [who works with a pharmaceutical company] struggled because the data set she was working with was incomplete -- 90% of the data set was male. They were trying to build a drug, which they'll sell to everybody, but they needed more data for a diverse population. So, she went out to get the data set to do the research.

If we don't have diversity now, it's going to hurt our future.

Proper research is when you have diversity in the data to get the right result.

How would you define diversity?

Saxena: Diversity is not just gender, color or ethnicity. It's also age. What you want is diversity on your team to get the right result.

We are building a digital world. You can do more commerce in a digital world. We are engaging more in a digital world, and we don't have enough people in the digital world. We don't have enough people who are coding, which is crazy. If we don't have diversity now, it's going to hurt our future. So, especially in digital data and AI, we need diversity more than anything else.

How can AI team leaders build diversity and build more responsible AI technology?

Saxena: Every leader needs to think about how they're building their teams and how they're looking at the data.

It's not enough that your teams are diverse. It's very important that the decision-makers -- the people who have the power to make changes -- are diverse. So, just giving token diversity is not enough. It's the equity that these women have the power to create through change. And women bring a different way of leadership than men do.

People are scared of AI. It is scary because one type of person is doing it. It won't be scary if we consciously build our teams to give power to the right people.

What is keeping some AI team leaders from pursuing diversity?

Saxena: I don't think they have a choice. It is scary for senior leaders right now. There's a huge McKinsey study [referring to this as] the 'Great Breakup.' Senior women leaders are leaving because there are microaggressions in organizations. So, you want your senior women to be strong.

Organizations need to invest in their senior leaders to make sure there's diversity on the top, not just in teams. Companies need to understand that having diversity on the top will make them money, but it will also make them look good. They'll be socially conscious. They'll innovate better; they'll produce better products. So it's a win-win.

How has the AI market changed negatively and positively for women?

Saxena: We've regressed a little bit. Even in conferences, we only have 15% to 20% attendance of women.

But how we've progressed is that AI demands diversity. AI will not succeed if you don't have diversity. If you want to be an innovative company, you want an edge over your competition. You want to be a technology leader. You want to be a data-driven company. You want to be an AI-driven company. You don't have a choice but to be diverse. And if you consciously start investing in it now rather than later, you will benefit. That's the progression that AI will force on you.

Editor's note: This interview has been edited for clarity and conciseness.