ai will take more jobs from women than men in the uk

While we’re busy fighting to end misogyny among humans and get paid the same as men, the robots are coming for our jobs, according to a new study.

by Clementine de Pressigny
|
18 April 2018, 1:06pm

Gloomy news from the immanent robot invasion now: machines may not be the bias-free blank slates we were be hoping for. Instead of mitigating the rampant sexism and racism of our human society, it turns out AI could, in fact, exacerbate it. Robots are programmed by people, and in this capitalist patriarchy we live in some people are awful, so sexism and racism can end up encoded into AI.

"It’s inevitable that AI will replace humans in much of the workforce -- current predictions are saying 800,000 million jobs will be affected."

This is worsened, of course, by the extreme lack of diversity among those working in tech. In 2016, just over a quarter of the computing workforce were women, and they are predominantly white. Women of colour account for less than 10%. At all levels, and particularly at the top levels of tech companies, white men are the vast majority, as they seem to be in all areas that involve lots of power and money. And as we know all too well, having those guys hold all the power creates more than a few issues. The industry is now well documented as being very hostile to women, sexual harassment is rampant, and people of colour are routinely discriminated against.

This is the backdrop to where we’re heading -- towards the second industrial revolution, predicted to kick off in the 2030s. Oliver Twist 2.0. The first industrial revolution changed everything about the way we live. The next one will do it all over again; some in ways that are just beginning to be predicted, some in ways we can’t know until they happen.

"Imagine a job in which your racist, misogynist colleague doesn’t just seem inhuman, but actually is."

We can’t completely freak out just yet, but it’s good to keep in mind that Elon Musk, lover of space and driverless cars, has stated that, “AI is a fundamental risk to the existence of human civilisation”. So, we should really be prepared. Because it’s inevitable that AI will replace humans in much of the workforce -- current predictions are saying 800,000 million jobs will be affected. Imagine a job in which your racist, misogynist colleague doesn’t just seem inhuman, but actually is. Just what you don’t need in the fight for equality, which we’re still engaged in with real people. Sure, management might seem like a bunch of robots, but what if they actually are robots. Well, that’s actually one thing you probably don’t have to worry about, because management are not the ones who will be losing their jobs to AI. No, that will be happening to those lower down the ladder. And according to a recent survey, in the UK it is women who will predominantly lose out compared to men when it comes to the robot job snatchers. For women in the US, the outlook is slightly less worrying, and that’s because they hold 15% of leadership roles in the workplace, compared to the UK where only 8% of director and managerial roles are held by women.

"The 2030s don’t have to be the decade that women lose out to discriminatory job-stealing bots."

In terms of particular industries globally, fashion could be hit particularly hard on a production level. Alex English, lead researcher behind the study, says that, "The research looks primarily at the technical expertise of the role as opposed to the industry, so while we cannot see at a glance how robots would affect for example, the fashion industry, we know a large proportion of roles here are in manufacture. These sorts of roles would fall under the bracket of 'technicians and mid-level professionals' or 'operators of facilities and machines and assemblers' -- both of which have an extremely high probability of automation (52% and 83% respectively) so we can ascertain that the outlook is perhaps not good for many people in this sector.” This could have positive impacts in some ways -- garment workers in countries like Bangladesh, primarily women, are terribly paid and work in dangerous environments. But being replaced by robots doesn’t mean those workers will get better jobs elsewhere.

However, the 2030s don’t have to be the decade that women lose out to discriminatory job-stealing bots. Tech entrepreneur and activist Tabitha Goldstaub says we can change that trajectory. “The PWC report shows that women are more likely to be worse affected by losing their tasks and jobs to automation in the short term,” she explains, “but long term this will level out as long as women are encouraged to leverage AI rather than having tech rule them.”

"Thankfully there are those doing research into unpicking discriminatory attitudes in humans, to help stop us building racist and sexist tech."

Which is exactly what Tabitha is working to ensure. “There is a lot of work being done to get more women into tech which is great, and it's also important that we focus on getting more women to work alongside tech,” she says. “Women are uniquely positioned to succeed in a world where we have to work alongside machines. The very personality traits traditionally encouraged in women, such as caring, empathy, patience and nurturing, which traditionally pointed to professions such as teaching and nursing are, in fact, the very characteristics which belong to the enlightened programmer AND are exactly what’s needed to work in any job alongside a machine.”

It’s also essential that we can recognise potential bias in order to eliminate it. Thankfully there are those doing research into unpicking discriminatory attitudes in humans, to help stop us building racist and sexist tech. Data science Professor James Zou has worked with researchers to study the way stereotyping takes hold and changes over time, by analysing word embeddings -- basically words that continually get linked together, like, say “woman” and “gentle”. James and his team looked at fiction and non-fiction texts from every decade of the 20th century and could see how “large cultural events like the women's movement and large immigration events do quite directly affect how people use certain words,” as James explains. Their research not only can provide a way to pinpoint and understand discriminatory language patterns, it can break these patterns. “The stereotypes have certain geometries in the data. Essentially if you modify the geometry a little bit you can remove these stereotypes,” James says. He explains that word embeddings are used in all sorts of machine learning AI applications, and so by using this information they can help stop robot bias, which is good news for women and minorities everywhere. Tabitha says it’s a constant battle to ensure companies creating AI are taking these factors into consideration. “We encourage corporates to use AI technology vendors that take transparency and accountability seriously and pick products that are tested for bias as well as having ethics committees that are involved in the deployment from the beginning.” The next frontier for the feminist fight is already being built.

Sign up for i-D's newsletter and get i-D straight to your inbox

Tagged:
feminism
robots
Artificial Intelligence
AI
SEXISM