Does Artificial Intelligence help or hinder the fight for gender equality? EIGE’s Director Carlien Scheele delved deeper into this digital dilemma and presented EIGE’s latest research on AI during a webinar held by the S&D Working Group on Artificial Intelligence and Gender Equality on 30 March 2022.
Artificial Intelligence has permeated almost all aspect of our lives.
And this is just the beginning.
While it may seem that AI and gender equality are not related, they are actually intimately intertwined.
The future of AI is female, a future which will be devoid of unconscious bias, a future where all voices are heard equally.
In this speech, I will explore what the future of AI looks like for gender equality in the workplace and beyond.
Did you like that introduction?
I had an AI write it.
The fact that most of you have not noticed shows how far artificial intelligence has come in the last years.
It was right about a few things. Artificial intelligence and gender equality are intimately intertwined.
This is because AI is a human creation. And humans are not perfect.
Let me walk you through some of these imperfections.
They start before AI is even developed.
Women are less likely to pursue science, technology, engineering and mathematics at school. They are often deterred by gender stereotypes.
The women who do make it to university face further gender segregation in male-dominated fields.
On graduation, some of these women will enter the field of AI.
Here they can face challenges such as male-dominated work environments, stereotyping, sexual harassment, and a lack of access to funding.
Is anyone surprised that only 16% of high-skilled AI workers are women?
Some of you may think: “So what?”. Women and men choose different career paths, and it is not for us to try to control this.
Yet the low numbers of women in AI are a symptom of a larger problem.
That problem is representation.
Representation is key to ensuring fair, equal, and democratic Artificial Intelligence. Until AI systems reflect our society in all its diversity, AI will cause more problems than it solves.
Because when we fail to account for gender, public spaces become male spaces by default.
Without representation, exclusively male experiences shape our modern reality. And as we’ve seen throughout history, those kinds of systems don’t tend to work so well for women.
Without representation, women miss out on well-paid and respected jobs in a growing industry. Professional sectors dominated by men have long had some of the best conditions in the labour market. This includes the AI sector.
Without representation, biased data sets used to train algorithms perpetuate stereotypes about gender.
Let me put this into context. A company trains an AI system to help with hiring. They train it using the CVs of existing employees. This makes sense so far.
However, this company is male-dominated. Their CVs do not tend have gaps for maternity leave. Their CVs do not mention being a member of a women’s sports club.
In fact, their CVs may not include the word “woman” at all. This lack of representation leads the system to penalise any CV that does not fit this restricted profile.
Unless we incentivise representation in industry and academia to increase the diversity of the AI workforce, design bias will continue to reproduce these gender stereotypes, as well as stereotypes about others not represented in the AI workforce, such as people of colour.
This is why I am happy the European Commission is taking action with its proposed Artificial Intelligence Act, which shows promise in minimising the risk of bias and discrimination in AI.
I also welcome the Commission’s commitment to train more women and people from diverse backgrounds to become AI professionals. Here I’d like to highlight the good work being done by the people behind initiatives such as ‘Women in AI’, who are helping propel more women into this sector, and who we must listen to and support as much as possible.
But there are also women who don’t work in the sector themselves, but whose work is nonetheless affected by AI.
I’ll take the example of work organised through platforms, or ‘gig work’ as it is often known. This growing sector includes food delivery and ride hailing services, such as Uber. 28 million Europeans have already done work via such platforms.
At EIGE, we interviewed some 5,000 of these workers to understand their motivations, and the challenges they face.
We found out that they are mainly young, highly educated, and taking care of children – in particular if they are women.
Indeed, more than a third of women doing platform work told us they do it because they can combine these jobs with family commitments and housework.
Yet algorithmic scheduling monitors platform workers, and can deduct ‘low productivity time’ from paid work time, leaving platform workers vulnerable to this digital wage theft.
A quarter of women platform workers are rarely able to work fixed schedules, or plan when and how much they will work.
This kind of software poses a particular threat to those caring for small children.
EU regulation is often out of step with technological developments. Policy frameworks must become future-proofed to respond to the challenges of new digital technologies.
If we do it right, there is much potential in these new areas of artificial intelligence.
For example, we found that in online platform work, there is a higher share of men doing jobs usually done by women than there is in the traditional labour market. This includes housework, childcare, and data entry.
Having more men being paid to take care of children and people’s homes can help chip away at stereotypes about women’s and men’s work, and of how much it is valued.
Without mainstreaming gender into all policy related to AI transformation of the labour market, we will lose these critical opportunities to advance gender equality.
I will leave you with a thought.
One of the very first representations of artificial intelligence came from a woman.
When Mary Shelley wrote Frankenstein in 1818, who would have predicted her story of a human made creature would remain relevant in 2022.
Frankenstein’s creation learned how to live from human input, watching a family through a hole in a cottage wall. Frankenstein’s creation was an empty shell. It learned its behaviour from humans. Mary Shelley showed us that Frankenstein’s creature itself was not monstrous, only the humans it observed.
200 years later, is the design of artificial intelligence much different to Mary Shelley’s conception?
The success of Artificial intelligence hinges upon how humans design, programme and use it.
And without greater representation, AI has the potential to hinder the fight for gender equality.
The gender perspective must be mainstreamed throughout all stages of AI design.
AI will then be able to offer an opportunity for us all.