Could gender bias become imbedded in AI technologies?

Could gender bias become imbedded in AI technologies?

Could the lack of women working within artificial intelligence (AI) have a detrimental effect on our society? Nikolas Kairinos, CEO and founder of Fountech, delves into the issue of gender bias in AI.

Casting a cursory glance at the discussions taking place in forums and the press, there is no shortage of debate on how artificial intelligence (AI) could have a negative impact on society – ranging from robots taking over human jobs, to existential concerns.

Of course, the vast majority of these concerns are more imaginative than credible, however it doesn’t mean the industry isn’t facing its own series of challenges.

One of the biggest challenges facing the AI industry is the gender diversity crisis. Namely, there are simply far too few women working in the field.

To shed some light on the gravity of the problem, researchers at AI Now Institute in April released a report that revealed how many women are involved in curating the future of AI.

It found that more than 80% of AI professors are men, while just 15% of AI research staff at Facebook and 10% at Google are women. 

These statistics highlight that we cannot become complacent and simply measure the success of the AI industry through the technological advancements made.

Instead, we need to tackle some of the key challenges holding the industry back from reaching its potential, promoting equality, diversity and opportunity for all.  

The dangerous feedback loop

Striving for diversity should by no means be limited to the tech and AI space; however, male dominance in a field that is responsible for the next generation of technological solutions risks perpetuating gender inequality within wider society.

This comes down to the simple fact that AI tools and technologies rely on humans to operate. Humans are responsible for writing the AI algorithms that power these technologies, after which the algorithms continue to feed on real-world data to inform their decision-making. As a result, they learn human biases and can start to mimic them.

The introduction of deeply ingrained bias can pose significant concerns for society, particularly given the increasing employment of AI solutions across all sectors.

To offer an illustrative example of how this might work in practice, Amazon recently came across this hurdle when it experimented with an AI-powered hiring tool.

The machine learning (ML) algorithms were trained to look for prospects by recognising terms that had appeared on the resumes of past successful job applicants.

While this might seem harmless enough, the tool eventually began downgrading resumes that contained mention of all-women’s colleges, or even application that simply contained the word “women’s”.

This just goes to show that if we are not careful enough, we risk exacerbating inequality in society through AI, rather than solving it.

What can we do?

Now that we understand the dangers posed by leaving the diversity problem left unsolved, what are some viable solutions that could help bridge the gender gap?

Importantly, the pressure of addressing this problem doesn’t just lie in organisations themselves.

The government must work together with the private sector to encourage more women to pursue careers in traditionally male-dominated fields like AI.

This necessarily starts with encouraging more women to seek out higher education in subjects like computer science and engineering, whether this is through grants or dedicated programmes.

As it stands, recent research by PwC recently revealed that only 15.8% of undergraduates in STEM fields across the UK are women.

It should come as no surprise then, that women are marginally represented in these professions.

Only 22% of AI professionals globally are female, compared to 78% who are male.

The greater burden must naturally fall on organisations themselves to address the issues preventing women from seeking a career in AI.

Positive steps have to be made to welcome more females into the workforce – particularly in higher-level positions where they can help drive change in the sector – and thereafter support them as they progress throughout the company.

For example, creating dedicated training programmes aimed at improving technical skills can go a long way towards giving women the confidence and ability to pursue more senior positions.

This could even entail creating a mentorship programme where women in senior positions guide junior employees towards advancement by offering personalised guidance and support.

More generally, organisations must think critically about why there are so few women in the field, and what they can do to change this.

I firmly believe that AI has the power to solve pressing global issues, but to achieve this, we must first stamp out the challenges holding the industry back, the most pressing of which is gender inequality.