By Teresa Harris
Early last year, former Uber CEO Travis Kalanick stepped down after several employees were ousted for their behaviour towards women, and the company was accused of fostering a toxic culture of sexism and harassment. In August 2017, James Damore, an engineer at Google, released a 10-page memo asserting women are biologically less suited to careers in tech, and criticized the company’s gender diversity efforts.
We’re familiar with these headlines, and the many others that have placed the state of gender diversity in North America’s tech industry under intense scrutiny. However, the problem goes deeper than we may realize — from the minds of employees, to the technology they’re producing — specifically in the realm of artificial intelligence. The result? Artificially intelligent technology that mimics the people and environment it was founded by and in: at best, inherently biased, and at worst, explicitly sexist.
Devices using artificial intelligence deeply affect how we live, work, and play. Voice-powered personal assistants are now with us in the car and kitchen, and suggestive search engines, which make use of machine-learning algorithms, seem to know us better than our closest companions. Since 2012, C.B. Insights reports that funding for A.I. start-ups has increased by over 850 per cent. Tech leaders including Google, Apple, and IBM have each purchased at least five companies with A.I. specialization, with Google acquiring a whopping 12 in the last six years.
“The problem goes deeper than we may realize — from the minds of employees, to the technology they’re producing.”
The consequences of the gender and racially homogeneous work environments characteristic of Silicon Valley are already being seen in the A.I. market, which comes as no surprise to many industry experts.
“When you don’t have the diversity of people designing voice-recognition software, you forget to test the technology using those people,” says Dr. Sarah Saska, co-founder and managing partner of Feminuity, a Toronto-based consulting firm that works with innovative companies to help them navigate through the unmapped territory of diversity, inclusion, and belonging. “Still to this day, some A.I. software doesn’t understand particular types of accents, i.e. those that deviate from the Western white male.”
University of Virginia computer science professor Vicente Ordóñez found that research-image collections supported by Microsoft and Facebook have shown “predictable gender bias in their depiction of activities such as cooking and sports,” strongly associating women with the former and men with the latter.
In 2017, news website Quartz studied how voice-powered assistants like Amazon’s Alexa, Apple’s Siri, and Microsoft’s Cortana responded to different types of verbal harassment, including lewd comments about their sex, sexuality, sexual characteristics, or sexual behavior. They found that “the bots most frequently evaded harassment, occasionally responded positively with either graciousness or flirtation, and rarely responded negatively,” meaning these virtual women almost never asked the harasser to stop, or told them that what they were saying was inappropriate.
These bots haven’t been around long enough to absorb the patriarchal biases entrenched throughout our society. However, the people — or should we say, the men — programming them have. And while bias and behaviour like this can be corrected, it requires a researcher to be looking for that bias in the first place, and to specify what he or she wants to correct. If recent headlines are any indicator, many within the tech industry don’t see the issue, or the value in correcting it.
“There are a lot of women who aren’t comfortable in environments where they don’t know everything. So encouraging them to take the leap is very important.”
Angelique Mohring is the founder and CEO of GainX, a company that uses A.I. and machine learning to aid global corporations in their transformation across people and projects. While she won’t deny the current state of gender inequality in the tech workplace, she remains hopeful that women not only belong there, but can add significant value to the field.
“Because of A.I., we’re going to need a skill set that goes beyond digital talent. The broader perspective women have will be worth its weight in gold in the future economy.” Mohring describes the ‘future economy’ as one wherein companies do much more with less — something she believes women are particularly well suited for. “Throughout history, women have always done more with less. We have been continually figuring out how to survive and take care of families and communities with very little.”
So how do we derail the speeding train that is biased artificial intelligence?
The obvious answer: get more women into tech so that more women, and a more diverse set of women, are designing and programming the tech we use. But Anne Martel, co-founder and SVP of Operations at Montreal startup Element AI, doesn’t think it’s as simple as getting more women in the door.
“It’s the company’s responsibility to be a safe place to learn, fail and learn from that failure,” she says. “There are a lot of women who aren’t comfortable in environments where they don’t know everything. So encouraging them to take the leap is very important.”
Even still, Martel thinks the consequences of non-diverse tech go beyond sexist and discriminatory software — she believes biased A.I. is destined to fail in the marketplace. “To allow for the adoption of A.I. systems, they have to be relatable. And a lack of diversity will prevent us from truly benefiting from these systems, because they’re not going to represent the reality we know.”
The One-sided State of Tech
According to data from the National Science Foundation, the number of women holding computer science degrees has declined from 25 per cent in 2004, to 18 per cent in 2014. And research from Morgan Stanley revealed that just 29 per cent of employees in tech are women, and only 13 per cent are executives.
The cause of a female shortage in tech comes down to what, in 2008, the Harvard Business Review called “The Athena Factor.” At the time, a reported 63 per cent of women in STEM (science, technology, engineering, and math) experienced sexual harassment at work, the result of cultures that celebrated “hostile machismo.” The review found other antigens that deter women from workplaces, including isolating them on teams of predominantly men and using systems of risk and reward that tend to disadvantage risk-averse women. The result? A 52 per cent drop off between women who graduate with degrees in STEM fields, and those who remain in those industries.
A more recent survey conducted in 2015 by a group of female tech investors and executives, titled “The Elephant in the Valley,” revealed that “84 per cent of the participants had been told they were too aggressive in the office, 66 per cent said that they had been excluded from important events because of their gender, and 60 per cent reported unwanted sexual advances in the workplace.”