Blog by Marie-Christine Gries, SI UN Representative in Paris.
On 8 March, International Women’s Day, UNESCO and the World Economic Forum organised a webinar focused on women in digital technologies, called “Girl Trouble: Breaking through the bias in AI”. Gabriella Ramos, Sustainable Development Goals sector for Human Social Sciences at UNESCO opened the session with this cry of alarm:
“The fourth industrial revolution is at our doorstep and gender equality risks falling by several decades”.
Today, all industries use new communication technologies. In the creation of these technologies, there is a continued need for the manufacturing of products and organisation of work. These technologies developed and implemented using artificial intelligence (AI) multiply the capabilities of industry, including scientific research in all fields. Many jobs are disappearing with the transformation of tasks by robotics, but at the same time, jobs are being created at a steady pace in the technological sectors, especially programming.
The statistical data on these sectors is worrying. Whilst young female students finally seem to be moving more towards scientific disciplines, very few integrate the sectors of AI technologies (only 24% of women in information science graduates). Women make up 22% of workers in this sector and only 6% in coding. Due to the major changes taking place in the skills required for employment, the future risks of unemployment pose a greater threat to women. Moreover, as the creation of innovative companies explodes, only 10% of startups are created by women (and not all rely on technology).
The second topic of this webinar was: “How can AI help us advance the rights of women and girls in society?” In other words, the inclusiveness of artificial intelligence. Gabriella Ramos declared:
“AI is not a reflection of bias, it is an amplifier”.
All discriminatory stereotypes and prejudices are present in algorithms, not just those that concern gender. Several studies cited by the panelists reveal its presence in the programming. Many AI applications are taking place today in everyday life: recruitment questionnaires, police investigations, online gaming programmes, facial recognition software, home device automation , advertising targeted algorithms, and virtual personal assistants often with the seemingly devoted voice of a woman. It is suggested that automated recruitment questionnaires have implicit biases, police software has the potential to exacerbate racial prejudice, some online games carry a wide range of discriminatory practices, and targeted advertising can lean on stereotypes. These applications that users install and use easily and freely can reinforce their prejudices and stereotypes. In many cases, the user does not even suspect the bias introduced by the algorithm. Even the name of Artificial Intelligence masks the reality of human intelligence at the source of inventions and applications.
Including more women in the sector and educating users on the realities of AI is imperative to make technology and digitalisation, inclusive.
Remedies cited at the end of the webinar were to:
- Push girls towards science, inspire mentorship;
- Include minority groups and people with disabilities;
- Include women at all levels of technology, especially programming;
- Introduce transparency in coding;
- Improve accountability of the technology sector and governments.
(Lead image from UNESCO)
You can watch the video recording of the webinar HERE.