The danger of a single story: Building AI without bias by ensuring more black women work in data
There is a great danger in “a single story”, says the novelist Chimamanda Adichie. If we hear only a single story about another person, group, culture or country, we risk critical misunderstandings.
For employers, and tech businesses, this is a pertinent issue – failure to mitigate against this risk could result in talent pouring out of an organisation, hindering its long-term growth.
This is a timely topic as the tech workforce undergoes an aggressive shake-up.
On the one hand, countless studies have shown the importance of diverse teams, with gender a common focal point. Noland and Moran (2016) published a study entitled, ‘Firms with More Women in the C-Suite Are More Profitable; the reason is that women think differently’, with research showing that large companies with at least three women in leading positions see a 66% increase in ROI.
However, analysis of tech layoffs in the US showed that 45% of those who lost their jobs in the recent wave of redundancies were women, despite the fact that women account for less than a third of tech industry workers. In other words, a disproportionately high number of women in tech lost their jobs in 2022. It would be interesting – and undoubtedly important – to see comparable data for the recent layoffs in UK tech.
Diminishing diversity enhances the danger of a single story. The challenge of achieving gender equality in the tech sector seems greater in the current climate, and where women of colour are concerned, the data is more worrying still.
A British Computer Society and Coding Black Females report – The Experiences of Black Women in the IT Industry, published in October 2022— found that just 0.7% of black women in the UK work in the IT industry, compared to 1.8% across the UK’s entire workforce. The study also revealed that women of all backgrounds and ethnicities make up around 22% of IT professionals (approximately 424,000), compared to 48% when examining the entire UK workforce. In other words: women, particularly black women, are far less likely to work in tech than other sectors.
Bias in data and AI
For more than a decade we have heard the call getting louder from businesses: “Data is the new oil”. Similarly, business adoption of artificial intelligence (AI) and data-intensive decision-making is becoming more and more common.
But set against the backdrop of so few women of colour working in UK tech, the emphasis on data and AI ought to ring alarm bells.
An increasing corpus of research reveals that racial and gender biases in machine learning algorithms and artificial intelligence-powered are partly caused by a lack of diversity in the data science community, which could escalate the existing inequities if unresolved.
“Machines can discriminate in harmful ways”, explains Joy Buolamwini. She points towards statistics highlighting the failures of AI to properly identify the gender of darker-skinned women compared to lighter-skinned women, before raising the serious concern that AI and ML can, if created by homogenous teams with natural bias, create “systems that amplify, rather than rectify, sexist hiring practices, racist criminal justice procedures, predatory advertising, and the spread of false information”.
Indeed, a 2021 Deloitte survey of women and men working in AI and machine learning (ML) found that the inclusion of women will bring unique perspectives to the development of advanced technology.
The rising chatter surrounding OpenAI and ChatGPT further highlights the importance of tackling these problems with great urgency; the proliferation of AI and data only makes the need more acute for greater diversity in the creation of AI and data solutions.
Getting more black women working in tech and data
Despite the major issues that lie before us, it is important to note that many outstanding black women are doing remarkable things in the data field and creating tables for upcoming black women to make giant strides. There has also been an upsurge of black-owned businesses tirelessly leading and promoting the preparation of women for corporations looking to embrace equity.
This is the mission of Niyo Group, which upskills black women for tech and data roles, and much more besides. Oyinkansola Adebayo, CEO and Founder of Niyo Group, explains: “It is imperative for black women to be included in the building of our data workforce to avoid bias from showing up in our data sets and algorithms, black women in data means black female representation in data.”
Olaoluwa Dada, COO of Niyo Group, adds: “Caroline Criado Perez once said ‘It’s not always easy to convince someone a need exists, if they don’t have that need themselves’. The people who build the products should look like the society they are building them for. So, when black women work in data and build AI-driven products, we are allowing them to be seen (by the data) and providing proof they exist in our world.”
So, how can we achieve this in practical terms?
The role of digital skills bootcamps
Digital skills bootcamps are a great example. On offer across the UK, the bootcamps are free of charge for learners and equip them with digital skills, giving them access to roles in areas like coding, cybersecurity and digital marketing.
Niyo runs skills bootcamps in collaboration with the West Midlands Combined Authority. They support the unemployed, those seeking a career change, as well as employed people looking to gain the digital skills required to secure more responsibility or a promotion with their current employer.
Niyo’s bootcamps are geared towards black women looking for a pathway into a tech career, giving them the chance to learn online and flexibly within a growing community of other black women in tech.
Back to Chimamanda’s narrative, AI bias is the real danger of a single story. Diversity in building these technologies will diversify the data used to build the algorithms; hence the urgent need for corporates to hire black women.
I would encourage all employers and those seeking new skills to search for digital skills bootcamps in their area. They have a significant role to play in facilitating digital skills training, while also creating career pathways for people irrespective of their background.
Chinazor Kalu is a data & AI subject-matter expert for the digital skills bootcamps run by Niyo Bootcamps. Niyo is a multifaceted organisation that uses innovative, creative tools such as technology, hair, and beauty to economically empower black women. With funding from the West Midlands Combined Authority, Niyo runs a number of bootcamps geared towards black women looking for a pathway into a tech career, giving them the chance to learn online and flexibly within a growing community of other black women in tech.