The challenges facing an expanding AI environment

For an industry at the forefront of the latest tech innovations and growth, is AI at risk of perpetuating gender inequality rather than solving it? Nikolas Kairinos, CEO, Fountech.ai, elaborates on this. 

---

Where developing new technology is concerned, the artificial intelligence (AI) industry is one of the most forward thinking and disruptive in the world today. Ambitious companies willing to operate at the cutting edge of AI and machine learning (ML) have the potential to fundamentally transform society by developing innovative solutions to the problems that plague everything from the way we treat diseases to the way we consume entertainment.

Despite this near limitless potential, AI companies are routinely criticised for their lack of diversity, particularly where gender is concerned. According to research commissioned by the World Economic Forum (WEF) only 22% of global AI professionals are women. At leading companies, the situation appears even worse as women comprise just 15% of AI research staff at Facebook and a mere 10% at Google.

These statistics are not simply food for thought. As AI is increasingly incorporated into a wider range of software and appliances, the industry requires greater gender diversity if it is to meet the needs of society at large.

The hidden danger of gender inequality in AI

Of course, gender imbalances are not limited to AI research; they are prevalent across the whole of the tech industry. However, there is a unique danger where AI and ML are concerned because algorithms depend for their functionality on access to relevant real-world data. Consequently, even well-designed AI tools have the potential to adopt and amplify the cognitive biases of their creators.

This might seem nebulous but the potential pitfalls are already evident in those areas where AI systems are being employed. For example, Amazon recently employed an experimental AI-driven hiring tool in order to more efficiently review resumes. However the program’s creators soon discovered that the algorithm was systematically discriminating against female candidates because they did not share the same interests or use the same language as previous successful applicants who were overwhelmingly male. The system even went as far as to penalise applicants for attending an all-women’s college or for including the word “women’s” as in “women’s chess club captain” in their resume.

What makes this an interesting case is that the algorithm was, if anything, functioning too well according to the narrow parameters for success that its developers had laid out for it. Like any technology, AI is not inherently flawed, rather it merely exacerbates pre-existing limitations within human thinking. As a result, diversifying the talent pool is not only laudable but necessary if we hope to mitigate against the possibility that AI will only provide a new means of discriminating against marginalised groups.

What is to be done?

Women are underrepresented in AI for several longstanding structural reasons and to suggest that this will change overnight does a disservice to the scale of the problem. However, in order to address the proliferation of gender biased AI, it is imperative that more women are brought into the industry.

The starting point for achieving this goal is making STEM degrees more attractive and accessible to women. This can be done directly through offering tailored grants or more subtly by giving more public exposure to the women currently pursuing careers in tech. Recent research from PwC reveals that only 15.8% of UK undergraduates in STEM fields are women and in reality, very little can be done by companies themselves to reverse the gender imbalance until there are a greater number of willing and able female candidates.

Retention also forms a crucial part of any diversity strategy and this is where AI companies can have the greatest impact. It is estimated that 56% of women who enter the tech industry leave before they attain mid-level jobs with many citing poor management and a lack of opportunity for career progression as reasons for their decision. Simple steps like establishing mentorship programmes designed to develop female leaders and implementing flexible work schedules can dramatically improve a company’s ability to attract and retain top female talent.

Ultimately, developing better AI doesn’t simply require a narrow focus on the technology itself but rather a more critical look at industry practises. AI has the power to change the world but for it to be truly transformative, the industry cannot continue allowing gender inequality to hold us back.