Original article was published by Santi Garcia on Artificial Intelligence on Medium
The Global Gender Gap Report 2018 recently published by the World Economic Forum devotes one of its sections to the worrying gender gap we observe in a professional field with one of the brightest futures: Artificial Intelligence (AI).
To elaborate this section of the report, the analysts of the World Economic Forum used data from LinkedIn, in particular information on the AI skills the users of this platform claim to possess. Of course, what a person says on her LinkedIn profile is not a guarantee that the person is a true expert in the subject. But still, the gender gap revealed by this data is very significant: only 22% of AI professionals in the (LinkedIn) world are women, compared to 78% men, which means that each woman who works in this field has an average of 3.5 male colleagues.
Another issue that draws the attention of the authors of the report is that although there are more and more LinkedIn users that include artificial intelligence skills in their profiles, the gender gap in this professional field has remained practically the same for the past four years.
Dimensioning the problem
By geography, the figures reveal that the gender gap in the field of artificial intelligence is a global problem. The gap in the two countries with the highest proportion of artificial intelligence specialists, United States and India, is around the average, but in the third country of the ranking, Germany, only 16% of AI professionals are women. In Spain, that ranks seventh in the list (according to what LinkedIn users say on their profiles), we are also below the average, with only 19% of female professionals. But even in countries where the gap is smaller (Singapore, Italy, South Africa) the presence of women in this domain does not exceed 28%.
From an industry perspective, the gender gap in AI is present in all industries, although it tends to be relatively smaller in those sectors where there is a greater presence of women, such as non-profit organizations, health services or education, and higher in more masculinized sectors such as manufacturing, energy, mining, hardware, software and IT services.
In regards the specific disciplines within AI in which men and women specialize, there are also some differences. Machine learning and data structures are the two most popular skills among both men and women in AI, but even here we see some differences: For example, 40% of female AI professionals say they posses machine learning skills, versus 47% of men. Deep learning, neural networks and artificial vision are other skills where the gender gap is greater than the average. But still, there are some capacities relatively more present among female AI experts than among their male colleagues such as information retrieval, natural language processing, and data structures.
It is also striking the difference between men and women in AI with regard the jobs and roles each of them play in organizations. Women are more likely to hold positions as data analysts, research, information management, and teaching (e.g., 4.2% of female AI professionals work as data analysts in contrast to 3.0% of men). On the contrary, men are better represented in roles such as Software Engineer, Head of Engineering, Head of IT, business owner and CEO, positions of higher hierarchical level and generally more lucrative. In other words, we face a gender gap that not only results in less women specialized in the field of artificial intelligence, but also in different professional trajectories for men and women that contribute to increasing the size and consequences of the gap.
The implications of the gender gap in the field of artificial intelligence are diverse and require urgent action for several reasons:
First, the gender gap in the field of artificial intelligence can exacerbate the gender gap in terms of economic participation and opportunity in the future as AI skills are increasingly demanded in the labor market. That is, the shortage of women in one of the professional disciplines with the brightest future (and the biggest pay checks), can have a negative impact on the average remuneration of women in the workplaces as well as their employment opportunities, since they lack of one of the skills sets that will be most demanded by employers in the near future.
Second, the gender gap in AI means that this technology, a ‘general purpose’ one, is being developed by not very diverse teams of technologists, which limits its innovative and inclusive capacity. In this regard, it is very significant the news that in early 2017 Amazon decided to scrap an AI solution to select new employees because it was not neutral with respect to the gender of the evaluated candidates. The problem was that the algorithm had been trained taking as a reference the profiles of the candidates the company had evaluated in the last 10 years, perpetuating the biases of human recruiters. As Margaret Boden pointed out a few months ago: perhaps with more women in this field, many of the problems AI tries to solve would be looked at differently and from a more empathetic perspective…
Third, the scarcity of female AI professionals, even in industries and geographies where there is a relatively high presence of women among technological talent, represents a missed opportunity in a domain characterized by a shortage of qualified personnel. Moreover, being a general purpose technology and, therefore, being in demand by most industries, the shortage of female talent in the field of artificial intelligence may cause that in the medium term the gender gap will widen, even in industries or geographies where there is no gender gap or this gap has been significanlty closed in the past years.
In conclusion, it becomes urgent to take action and implement measures aimed at closing the gender gap in AI. There are very interesting initiatives that point in this direction, such as AI4All, or AllWomen (an artificial intelligence campus only for women we had the opportunity to know first-hand on the occasion of the Social Impact Day we organized in Future for Work Institute). But this is not enough in front of such a big challenge. We need to do more…