Tackling the gender gap in AI - Gavriella Schuster

In this opinion piece, the former Microsoft global channel chief, and chairwoman of the advisory board at Artificial Solutions, explains what more can be done on this issue

Around 50 percent of the population are women. Yet women account for just 39 percent of the global workforce, and only 19 percent of tech workers.

Apple's chief executive Tim Cook recently commented on this, saying there are "still not enough women at the table" at the world's tech firms - including his own - and that there were "no good excuses" for this.

Artificial intelligence (AI) is one of the leading areas of investment in technology innovation today, and it too suffers from a huge gender disparity.

According to the World Economic Forum, only 22 percent of all AI professionals are women, and just 12 percent of machine learning researchers.

This is particularly troublesome for the technology's direction of travel.

Biased data

The aim of AI is to effectively augment the human experience and human capabilities. To do so, it must be designed by default to represent all humans. AI systems learn to make decisions based on the data with which they are trained and coded.

But this data can often be contaminated by the implicit biases of the humans that created it. This tainted data can then - unconsciously - reflect race and gender bias or other social inequalities.

It's clear, then, that if women are not sufficiently active participants in the development of AI algorithms, building the context of conversational AI, and testing the use cases, the male experience will become the default.

It is absolutely critical, therefore, to ensure gender balance in all aspects of the research and development of this emerging capability.

Valuing contributions

One of the reasons so few women are active in the field of AI is the way the subject is taught in academia, which has failed to put forward use cases for the technology that would attract women to the profession.

Focusing on the collaboration, human interaction, and social justice issues that would appeal to women, for example, would encourage more female participation in the profession.

Then, even once this has been addressed, and women are engaged, it's vital that they feel that their contributions are valued. But there's an unconscious bias that persists towards women that tends to drown out their voices, that sees them interrupted four times more frequently than their male counterparts, and that supports a ‘hero mentality'.

This way of thinking must be overcome to create a stronger environment of inclusion that values greater diversity.

Network of allies

Quite simply, the AI industry needs to make a concerted effort to recruit more women into the field, and to recognise and reward the contributions that women make.

As part of this, it needs to work on allyship as a way of creating a more welcoming culture for women to bring their whole selves and their best ideas into the industry.

Most men actually want to be allies; many, however, might not be aware of the number of micro-aggressions that women face on a daily basis.

They won't have noticed, for example, that a woman will be interrupted more often than her male colleagues, or that something she says will be repeated, only for someone else to take credit for it. In this context, an ally is someone that will support a woman the next time she's interrupted or ignored.

Creating a network of personal allies within a working environment is critical for creating a strong culture that is both productive and supportive of diverse talent. Artificial Solutions' AiAllies initiative, for example, is designed to help create such a network for women to connect with and learn from.

For someone to be an ally to a colleague that may need personal and professional support requires them to recognise that they have privilege and a degree of power with an organisation, and that they're willing to work with intention to confer that power to those around them. Essentially, an ally is someone who shares their power.

Expanded universe

Recruiters must step out of their comfort zones, too. A company's pipeline - the number of CVs it's receiving - will often depend on who its employees feel comfortable networking with. Employees need to take intentional steps to connect with people from outside their network, from different backgrounds and with different histories.

Expanding the range of its universe in this way, will increasing the diversity of a company's pipeline and, in turn, the make-up of its workforce.

It's important for women in AI, as in any other industry, to have networks of allies and other women to help them feel that they belong.

They need to hear that they're not alone in their struggles, and that they're not crazy for feeling as if they're often interrupted, overlooked, or dismissed. But, given the potential for bias in the data used to train the various systems and algorithms, there's perhaps an especially urgent need to address the issue in the AI industry right now.

Tim Cook was right, there is no excuse for the shortage of women in tech companies.

So, it's time that employers focused not only on recruiting more women, but also ensuring their contributions are heard and valued as much as their male colleagues.