Interview | Press corps
2 May 2024 | Although the industry needs workers and the development of responsible artificial intelligence benefits from the increase in the number of female IT professionals, both the industry and artificial intelligence training are still dominated by men. Lecturer Maaike Harbers (Rotterdam University of Applied Sciences) explains why female students feel less at home in artificial intelligence courses and outlines how cultural change can be initiated. “It requires different efforts than putting a photo of a female student on a brochure.”
According to a recent study, the development of responsible artificial intelligence is hampered by the lack of diversity in development teams. Led by Maaike Harbers, lecturer in artificial intelligence and society at the Rotterdam University of Applied Sciences, the researchers examined the development of responsible artificial intelligence and the role of women in it.
According to the researchers, artificial intelligence is a technology “at high risk of adverse ethical consequences”; Consider discrimination due to data bias or the role of algorithms in a benefits scandal. “It seems likely that more diversity in the AI sector will provide broader perspectives when designing AI applications and that this will reduce the potential for unintended consequences of AI applications,” the report says.
However, the ICT industry and the artificial intelligence industry are dominated by men according to different figures. In 2022, only fourteen percent of Dutch ICT workers were women. Not only is this detrimental to the diversity of viewpoints in AI development, but it also means that a large potential workforce is going untapped in an industry suffering from a growing labor shortage.
There are far fewer female students in ICT studies
It seems that education can only partially solve this problem. Women choose ICT studies much less frequently. In the academic year 2021-2022, only nine percent of university students in computer science were women. In university education, a quarter of computer science students were women. In addition, only slightly more than a quarter of women with an IT degree work in the ICT sector, while this is more than half of men.
Interviews with nineteen Dutch women in the AI sector show that “many women are systematically not taken or treated less seriously in the workplace. (…) In the worst case, it leads to women leaving the AI sector and thus a loss of talent in this sector. About 50 percent of female employees in the IT sector leave ten within a year.
Women recognize ethical problems better than men
However, women recognize the ethical problems related to artificial intelligence better than men, participate in them more often and pay more attention to them. In addition, the researchers point out that responsible AI receives too little attention in many companies in the Dutch AI sector. “Attention to this has grown rapidly in recent years, but it mainly concerns the technology itself – for example, bias in the data,” says lecturer Maaike Harbers in a conversation with ScienceGuide.
You can get the weekly newsletter for free for a while. You can find the terms and conditions here.
“However, diversity concerns the conditions surrounding it: who is responsible for responsible artificial intelligence in the organization? At what stage in the process is this considered? What are the procedures for this? Some companies have this arrangement, but most do not.”
This leads to a difficult situation, as the report shows: “responsible AI does not always get the attention it deserves, but a group that is relatively more committed to responsible AI and can make an important contribution by providing an additional perspective is the group that is taken less seriously in the field,” the researchers write.
Programming confers prestige, and women do it less
Artificial intelligence training should therefore focus on attracting and keeping women in the field by ensuring inclusive training, is one of the recommendations. Many courses, departments and institutions have already invested in this, for example by organizing information and speed dates with role models. “However, I think a lot more needs to be done,” says Harbers. “The programs today mainly focus on attracting female students, not so much on keeping them. It requires different efforts than putting a photo of a female student in a brochure.”
According to Harbers, female students feel less at home in an artificial intelligence or ICT course than male students. The reasons for this are mainly at the level of culture. For example, examples used in class are more likely to match the interests of male students. In addition, programming is something that students value each other – and boys starting the course often have more experience with it than girls. Male students also program more often in their free time than female students.
“It’s not necessary, right? You can easily study computer science by doing school assignments, but the courses often have a culture where programming as a hobby is seen as a good thing,” says the Rotterdam lecturer. “If it appeals to you less, you’re less likely to feel at home in a course like this.”
Female students are more likely to seek social meaning
Teachers can play an important role in this. For example, female students are more likely to be interested in socially relevant programming applications. By using such examples or tasks more often, a female student can feel more at home in the course, Harbers gives as an example.
“I once heard a teacher say at an open day at Rotterdam University of Applied Sciences that you are more likely to succeed if you already have programming experience. In addition to research support, this statement also lacks people with less programming experience—including girls.
The role of diversity must also be reflected in the content of education, Harbers emphasizes. “This can be done by making students aware that they experience the world from a particular perspective – shaped by upbringing, culture, education, gender, ethnicity, etc. – and that this affects the AI systems they design and develop. In order to develop responsible AI that works for everyone, it is therefore important to include multiple perspectives in the design and development process. By taking in the knowledge and experiences of others, you can compensate for blind spots from your own perspective.
It starts with our toys
The necessary cultural change actually goes much further than just artificial intelligence training, Harbers understands. “It starts with what kind of toys we offer the children. Boys are more often given toys that promote spatial vision, girls are encouraged more for communication and social interaction. In addition, boys and girls are approached differently in science subjects. A boy with a six in math is told to work a little harder, a girl with the same grade is asked if she should drop math.” It’s not easy to change, the Rotterdam lecturer understands. “However, that’s no reason to leave it at that.”